diff --git a/docs/build/html/_sources/dev/custom_metal_kernels.rst b/docs/build/html/_sources/dev/custom_metal_kernels.rst index c4c1b0aff..3e92f2814 100644 --- a/docs/build/html/_sources/dev/custom_metal_kernels.rst +++ b/docs/build/html/_sources/dev/custom_metal_kernels.rst @@ -1,3 +1,5 @@ +.. _custom_metal_kernels: + Custom Metal Kernels ==================== @@ -76,6 +78,10 @@ Putting this all together, the generated function signature for ``myexp`` is as template [[host_name("custom_kernel_myexp_float")]] [[kernel]] decltype(custom_kernel_myexp_float) custom_kernel_myexp_float; +Note: ``grid`` and ``threadgroup`` are parameters to the Metal `dispatchThreads `_ function. +This means we will launch ``mx.prod(grid)`` threads, subdivided into ``threadgroup`` size threadgroups. +For optimal performance, each thread group dimension should be less than or equal to the corresponding grid dimension. + Passing ``verbose=True`` to ``mx.fast.metal_kernel.__call__`` will print the generated code for debugging purposes. Using Shape/Strides diff --git a/docs/build/html/_sources/usage/function_transforms.rst b/docs/build/html/_sources/usage/function_transforms.rst index 9a15bbf1c..9769fceaa 100644 --- a/docs/build/html/_sources/usage/function_transforms.rst +++ b/docs/build/html/_sources/usage/function_transforms.rst @@ -161,7 +161,7 @@ A naive way to add the elements from two sets of vectors is with a loop: ys = mx.random.uniform(shape=(100, 4096)) def naive_add(xs, ys): - return [xs[i] + ys[:, i] for i in range(xs.shape[1])] + return [xs[i] + ys[:, i] for i in range(xs.shape[0])] Instead you can use :func:`vmap` to automatically vectorize the addition: @@ -169,7 +169,7 @@ Instead you can use :func:`vmap` to automatically vectorize the addition: # Vectorize over the second dimension of x and the # first dimension of y - vmap_add = mx.vmap(lambda x, y: x + y, in_axes=(1, 0)) + vmap_add = mx.vmap(lambda x, y: x + y, in_axes=(0, 1)) The ``in_axes`` parameter can be used to specify which dimensions of the corresponding input to vectorize over. Similarly, use ``out_axes`` to specify diff --git a/docs/build/html/_sources/usage/indexing.rst b/docs/build/html/_sources/usage/indexing.rst index 62994a0fb..c74e357fa 100644 --- a/docs/build/html/_sources/usage/indexing.rst +++ b/docs/build/html/_sources/usage/indexing.rst @@ -77,7 +77,7 @@ from the GPU. Performing bounds checking for array indices before launching the kernel would be extremely inefficient. Indexing with boolean masks is something that MLX may support in the future. In -general, MLX has limited support for operations for which outputs +general, MLX has limited support for operations for which output *shapes* are dependent on input *data*. Other examples of these types of operations which MLX does not yet support include :func:`numpy.nonzero` and the single input version of :func:`numpy.where`. diff --git a/docs/build/html/_sources/usage/lazy_evaluation.rst b/docs/build/html/_sources/usage/lazy_evaluation.rst index 466edaaed..8fd855efa 100644 --- a/docs/build/html/_sources/usage/lazy_evaluation.rst +++ b/docs/build/html/_sources/usage/lazy_evaluation.rst @@ -109,7 +109,7 @@ Here is a concrete example: An important behavior to be aware of is when the graph will be implicitly evaluated. Anytime you ``print`` an array, convert it to an -:obj:`numpy.ndarray`, or otherwise access it's memory via :obj:`memoryview`, +:obj:`numpy.ndarray`, or otherwise access its memory via :obj:`memoryview`, the graph will be evaluated. Saving arrays via :func:`save` (or any other MLX saving functions) will also evaluate the array. diff --git a/docs/build/html/backend_2metal_2device_8h_source.html b/docs/build/html/backend_2metal_2device_8h_source.html index 8b936bd40..bb0b3537b 100644 --- a/docs/build/html/backend_2metal_2device_8h_source.html +++ b/docs/build/html/backend_2metal_2device_8h_source.html @@ -149,7 +149,7 @@ $(function(){ initResizable(false); });
50 ~ConcurrentContext() {
51 enc.concurrent_ = false;
-
52 enc.outputs_.insert(
+
52 enc.prev_outputs_.insert(
53 enc.concurrent_outputs_.begin(), enc.concurrent_outputs_.end());
54 enc.concurrent_outputs_.clear();
55 }
@@ -170,212 +170,215 @@ $(function(){ initResizable(false); });
66 void set_output_array(array& a, int idx, int64_t offset = 0);
67 void dispatchThreadgroups(MTL::Size grid_dims, MTL::Size group_dims);
68 void dispatchThreads(MTL::Size grid_dims, MTL::Size group_dims);
-
69
-
-
70 ConcurrentContext start_concurrent() {
-
71 return ConcurrentContext(*this);
-
72 }
+
69 void maybeInsertBarrier();
+
70
+
+
71 ConcurrentContext start_concurrent() {
+
72 return ConcurrentContext(*this);
+
73 }
-
73 ~CommandEncoder();
-
74
-
75 // Inputs to all kernels in the encoder including temporaries
-
-
76 std::unordered_set<const void*>& inputs() {
-
77 return all_inputs_;
-
78 };
+
74 ~CommandEncoder();
+
75
+
76 // Inputs to all kernels in the encoder including temporaries
+
+
77 std::unordered_set<const void*>& inputs() {
+
78 return all_inputs_;
+
79 };
-
79
-
80 // Outputs of all kernels in the encoder including temporaries
-
-
81 std::unordered_set<const void*> outputs() {
-
82 return all_outputs_;
-
83 };
+
80
+
81 // Outputs of all kernels in the encoder including temporaries
+
+
82 std::unordered_set<const void*> outputs() {
+
83 return all_outputs_;
+
84 };
-
84
-
85 private:
-
86 MTL::ComputeCommandEncoder* enc_;
-
87 bool concurrent_{false};
-
88 std::unordered_set<MTL::Resource*> outputs_;
-
89 std::unordered_set<MTL::Resource*> concurrent_outputs_;
-
90 std::unordered_set<const void*> all_inputs_;
-
91 std::unordered_set<const void*> all_outputs_;
-
92};
+
85
+
86 private:
+
87 MTL::ComputeCommandEncoder* enc_;
+
88 bool needs_barrier_{false};
+
89 bool concurrent_{false};
+
90 std::unordered_set<MTL::Resource*> prev_outputs_;
+
91 std::unordered_set<MTL::Resource*> next_outputs_;
+
92 std::unordered_set<MTL::Resource*> concurrent_outputs_;
+
93 std::unordered_set<const void*> all_inputs_;
+
94 std::unordered_set<const void*> all_outputs_;
+
95};
-
93
-
-
94struct Fence {
-
95 Fence(MTL::Fence* fence) : fence(fence) {}
-
-
96 ~Fence() {
-
97 fence->release();
-
98 }
+
96
+
+
97struct Fence {
+
98 Fence(MTL::Fence* fence) : fence(fence) {}
+
+
99 ~Fence() {
+
100 fence->release();
+
101 }
-
99 MTL::Fence* fence;
-
100};
+
102 MTL::Fence* fence;
+
103};
-
101
-
-
102struct DeviceStream {
-
103 DeviceStream(MTL::CommandQueue* queue) : queue(queue) {};
-
-
104 ~DeviceStream() {
-
105 queue->release();
-
106 if (buffer != nullptr) {
-
107 buffer->release();
-
108 }
-
109 };
+
104
+
+
105struct DeviceStream {
+
106 DeviceStream(MTL::CommandQueue* queue) : queue(queue) {};
+
+
107 ~DeviceStream() {
+
108 queue->release();
+
109 if (buffer != nullptr) {
+
110 buffer->release();
+
111 }
+
112 };
-
110 MTL::CommandQueue* queue;
-
111 // A map of prior command encoder outputs to their corresponding fence
-
112 std::unordered_map<const void*, std::shared_ptr<Fence>> outputs;
-
113 // Used to allow thread-safe access to the outputs map
-
114 std::mutex fence_mtx;
-
115
-
116 // The buffer and buffer op count are updated
-
117 // between command buffers
-
118 MTL::CommandBuffer* buffer{nullptr};
-
119 int buffer_ops{0};
-
120
-
121 // The command encoder, fence, and temporaries are updated between command
-
122 // encoders
-
123 std::unique_ptr<CommandEncoder> encoder{nullptr};
-
124 std::shared_ptr<Fence> fence;
-
125 std::vector<array> temporaries;
-
126};
+
113 MTL::CommandQueue* queue;
+
114 // A map of prior command encoder outputs to their corresponding fence
+
115 std::unordered_map<const void*, std::shared_ptr<Fence>> outputs;
+
116 // Used to allow thread-safe access to the outputs map
+
117 std::mutex fence_mtx;
+
118
+
119 // The buffer and buffer op count are updated
+
120 // between command buffers
+
121 MTL::CommandBuffer* buffer{nullptr};
+
122 int buffer_ops{0};
+
123
+
124 // The command encoder, fence, and temporaries are updated between command
+
125 // encoders
+
126 std::unique_ptr<CommandEncoder> encoder{nullptr};
+
127 std::shared_ptr<Fence> fence;
+
128 std::vector<array> temporaries;
+
129};
-
127
-
-
128class Device {
-
129 public:
-
130 Device();
-
131 Device(const Device&) = delete;
-
132 Device& operator=(const Device&) = delete;
-
133 ~Device();
-
134
-
-
135 MTL::Device* mtl_device() {
-
136 return device_;
-
137 };
+
130
+
+
131class Device {
+
132 public:
+
133 Device();
+
134 Device(const Device&) = delete;
+
135 Device& operator=(const Device&) = delete;
+
136 ~Device();
+
137
+
+
138 MTL::Device* mtl_device() {
+
139 return device_;
+
140 };
-
138
-
-
139 const std::string& get_architecture() {
-
140 return arch_;
-
141 }
+
141
+
+
142 const std::string& get_architecture() {
+
143 return arch_;
+
144 }
-
142
-
143 void new_queue(int index);
-
144 MTL::CommandBuffer* get_command_buffer(int index);
-
145 int get_command_buffer_ops(int index);
-
146 void increment_command_buffer_ops(int index);
-
147 void commit_command_buffer(int index);
-
148 CommandEncoder& get_command_encoder(int index);
-
149 void end_encoding(int index);
-
150
-
151 void register_library(
-
152 const std::string& lib_name,
-
153 const std::string& lib_path);
-
154
-
155 // Note, this should remain in the header so that it is not dynamically
-
156 // linked
-
-
157 void register_library(const std::string& lib_name) {
-
158 if (auto it = library_map_.find(lib_name); it == library_map_.end()) {
-
159 register_library(lib_name, get_colocated_mtllib_path(lib_name));
-
160 }
-
161 }
+
145
+
146 void new_queue(int index);
+
147 MTL::CommandBuffer* get_command_buffer(int index);
+
148 int get_command_buffer_ops(int index);
+
149 void increment_command_buffer_ops(int index);
+
150 void commit_command_buffer(int index);
+
151 CommandEncoder& get_command_encoder(int index);
+
152 void end_encoding(int index);
+
153
+
154 void register_library(
+
155 const std::string& lib_name,
+
156 const std::string& lib_path);
+
157
+
158 // Note, this should remain in the header so that it is not dynamically
+
159 // linked
+
+
160 void register_library(const std::string& lib_name) {
+
161 if (auto it = library_map_.find(lib_name); it == library_map_.end()) {
+
162 register_library(lib_name, get_colocated_mtllib_path(lib_name));
+
163 }
+
164 }
-
162
-
163 MTL::Library* get_library(
-
164 const std::string& name,
-
165 const std::function<std::string(void)>& builder);
-
166
-
167 MTL::ComputePipelineState* get_kernel(
-
168 const std::string& base_name,
-
169 MTL::Library* mtl_lib,
-
170 const std::string& hash_name = "",
-
171 const MTLFCList& func_consts = {},
-
172 const std::vector<MTL::Function*>& linked_functions = {});
-
173
-
174 MTL::ComputePipelineState* get_kernel(
-
175 const std::string& base_name,
-
176 const std::string& lib_name = "mlx",
-
177 const std::string& hash_name = "",
-
178 const MTLFCList& func_consts = {},
-
179 const std::vector<MTL::Function*>& linked_functions = {});
-
180
-
181 MTL::ArgumentEncoder* argument_encoder(
-
182 const std::vector<MTL::ArgumentDescriptor*>& arg_descs) const;
+
165
+
166 MTL::Library* get_library(
+
167 const std::string& name,
+
168 const std::function<std::string(void)>& builder);
+
169
+
170 MTL::ComputePipelineState* get_kernel(
+
171 const std::string& base_name,
+
172 MTL::Library* mtl_lib,
+
173 const std::string& hash_name = "",
+
174 const MTLFCList& func_consts = {},
+
175 const std::vector<MTL::Function*>& linked_functions = {});
+
176
+
177 MTL::ComputePipelineState* get_kernel(
+
178 const std::string& base_name,
+
179 const std::string& lib_name = "mlx",
+
180 const std::string& hash_name = "",
+
181 const MTLFCList& func_consts = {},
+
182 const std::vector<MTL::Function*>& linked_functions = {});
183
-
184 // Record temporary arrays for the given stream index
-
185 void add_temporary(array arr, int index);
-
186 void add_temporaries(std::vector<array> arrays, int index);
-
187
-
188 void set_residency_set(const MTL::ResidencySet* residency_set);
-
189
-
190 private:
-
191 DeviceStream& get_stream_(int index) {
-
192 return stream_map_.find(index)->second;
-
193 }
-
194 MTL::Library* get_library_cache_(const std::string& name);
-
195
-
196 MTL::Library* get_library_(const std::string& name);
-
197 MTL::Library* build_library_(const std::string& source_string);
+
184 MTL::ArgumentEncoder* argument_encoder(
+
185 const std::vector<MTL::ArgumentDescriptor*>& arg_descs) const;
+
186
+
187 // Record temporary arrays for the given stream index
+
188 void add_temporary(array arr, int index);
+
189 void add_temporaries(std::vector<array> arrays, int index);
+
190
+
191 void set_residency_set(const MTL::ResidencySet* residency_set);
+
192
+
193 private:
+
194 DeviceStream& get_stream_(int index) {
+
195 return stream_map_.find(index)->second;
+
196 }
+
197 MTL::Library* get_library_cache_(const std::string& name);
198
-
199 MTL::Function* get_function_(const std::string& name, MTL::Library* mtl_lib);
-
200
-
201 MTL::Function* get_function_(
-
202 const std::string& name,
-
203 const std::string& specialized_name,
-
204 const MTLFCList& func_consts,
-
205 MTL::Library* mtl_lib);
-
206
-
207 MTL::LinkedFunctions* get_linked_functions_(
-
208 const std::vector<MTL::Function*>& funcs);
+
199 MTL::Library* get_library_(const std::string& name);
+
200 MTL::Library* build_library_(const std::string& source_string);
+
201
+
202 MTL::Function* get_function_(const std::string& name, MTL::Library* mtl_lib);
+
203
+
204 MTL::Function* get_function_(
+
205 const std::string& name,
+
206 const std::string& specialized_name,
+
207 const MTLFCList& func_consts,
+
208 MTL::Library* mtl_lib);
209
-
210 MTL::ComputePipelineState* get_kernel_(
-
211 const std::string& name,
-
212 const MTL::Function* mtl_function);
-
213
-
214 MTL::ComputePipelineState* get_kernel_(
-
215 const std::string& name,
-
216 const MTL::Function* mtl_function,
-
217 const MTL::LinkedFunctions* linked_functions);
-
218
-
219 MTL::ComputePipelineState* get_kernel_(
-
220 const std::string& base_name,
-
221 MTL::Library* mtl_lib,
-
222 const std::string& hash_name,
-
223 const MTLFCList& func_consts = {},
-
224 const std::vector<MTL::Function*>& linked_functions = {});
-
225
-
226 MTL::Device* device_;
-
227 std::unordered_map<int32_t, DeviceStream> stream_map_;
+
210 MTL::LinkedFunctions* get_linked_functions_(
+
211 const std::vector<MTL::Function*>& funcs);
+
212
+
213 MTL::ComputePipelineState* get_kernel_(
+
214 const std::string& name,
+
215 const MTL::Function* mtl_function);
+
216
+
217 MTL::ComputePipelineState* get_kernel_(
+
218 const std::string& name,
+
219 const MTL::Function* mtl_function,
+
220 const MTL::LinkedFunctions* linked_functions);
+
221
+
222 MTL::ComputePipelineState* get_kernel_(
+
223 const std::string& base_name,
+
224 MTL::Library* mtl_lib,
+
225 const std::string& hash_name,
+
226 const MTLFCList& func_consts = {},
+
227 const std::vector<MTL::Function*>& linked_functions = {});
228
-
229 std::shared_mutex kernel_mtx_;
-
230 std::unordered_map<std::string, MTL::ComputePipelineState*> kernel_map_;
+
229 MTL::Device* device_;
+
230 std::unordered_map<int32_t, DeviceStream> stream_map_;
231
-
232 std::shared_mutex library_mtx_;
-
233 std::unordered_map<std::string, MTL::Library*> library_map_;
-
234 const MTL::ResidencySet* residency_set_{nullptr};
-
235 std::string arch_;
-
236};
+
232 std::shared_mutex kernel_mtx_;
+
233 std::unordered_map<std::string, MTL::ComputePipelineState*> kernel_map_;
+
234
+
235 std::shared_mutex library_mtx_;
+
236 std::unordered_map<std::string, MTL::Library*> library_map_;
+
237 const MTL::ResidencySet* residency_set_{nullptr};
+
238 std::string arch_;
+
239};
-
237
-
238Device& device(mlx::core::Device);
-
239
-
240} // namespace mlx::core::metal
+
240
+
241Device& device(mlx::core::Device);
+
242
+
243} // namespace mlx::core::metal
mlx::core::array
Definition array.h:20
-
mlx::core::metal::Device
Definition device.h:128
+
mlx::core::metal::Device
Definition device.h:131
mlx::core::metal::Device::set_residency_set
void set_residency_set(const MTL::ResidencySet *residency_set)
mlx::core::metal::Device::get_command_buffer_ops
int get_command_buffer_ops(int index)
-
mlx::core::metal::Device::mtl_device
MTL::Device * mtl_device()
Definition device.h:135
+
mlx::core::metal::Device::mtl_device
MTL::Device * mtl_device()
Definition device.h:138
mlx::core::metal::Device::register_library
void register_library(const std::string &lib_name, const std::string &lib_path)
mlx::core::metal::Device::get_command_buffer
MTL::CommandBuffer * get_command_buffer(int index)
mlx::core::metal::Device::end_encoding
void end_encoding(int index)
-
mlx::core::metal::Device::get_architecture
const std::string & get_architecture()
Definition device.h:139
+
mlx::core::metal::Device::get_architecture
const std::string & get_architecture()
Definition device.h:142
mlx::core::metal::Device::get_kernel
MTL::ComputePipelineState * get_kernel(const std::string &base_name, MTL::Library *mtl_lib, const std::string &hash_name="", const MTLFCList &func_consts={}, const std::vector< MTL::Function * > &linked_functions={})
mlx::core::metal::Device::argument_encoder
MTL::ArgumentEncoder * argument_encoder(const std::vector< MTL::ArgumentDescriptor * > &arg_descs) const
mlx::core::metal::Device::add_temporaries
void add_temporaries(std::vector< array > arrays, int index)
@@ -383,7 +386,7 @@ $(function(){ initResizable(false); });
mlx::core::metal::Device::increment_command_buffer_ops
void increment_command_buffer_ops(int index)
mlx::core::metal::Device::new_queue
void new_queue(int index)
mlx::core::metal::Device::commit_command_buffer
void commit_command_buffer(int index)
-
mlx::core::metal::Device::register_library
void register_library(const std::string &lib_name)
Definition device.h:157
+
mlx::core::metal::Device::register_library
void register_library(const std::string &lib_name)
Definition device.h:160
mlx::core::metal::Device::Device
Device(const Device &)=delete
mlx::core::metal::Device::add_temporary
void add_temporary(array arr, int index)
mlx::core::metal::Device::operator=
Device & operator=(const Device &)=delete
@@ -402,31 +405,32 @@ $(function(){ initResizable(false); });
mlx::core::metal::CommandEncoder
Definition device.h:41
mlx::core::metal::CommandEncoder::dispatchThreads
void dispatchThreads(MTL::Size grid_dims, MTL::Size group_dims)
mlx::core::metal::CommandEncoder::CommandEncoder
CommandEncoder(MTL::CommandBuffer *cbuf)
-
mlx::core::metal::CommandEncoder::inputs
std::unordered_set< const void * > & inputs()
Definition device.h:76
+
mlx::core::metal::CommandEncoder::inputs
std::unordered_set< const void * > & inputs()
Definition device.h:77
mlx::core::metal::CommandEncoder::operator=
CommandEncoder & operator=(const CommandEncoder &)=delete
-
mlx::core::metal::CommandEncoder::start_concurrent
ConcurrentContext start_concurrent()
Definition device.h:70
+
mlx::core::metal::CommandEncoder::start_concurrent
ConcurrentContext start_concurrent()
Definition device.h:71
mlx::core::metal::CommandEncoder::set_output_array
void set_output_array(array &a, int idx, int64_t offset=0)
mlx::core::metal::CommandEncoder::dispatchThreadgroups
void dispatchThreadgroups(MTL::Size grid_dims, MTL::Size group_dims)
mlx::core::metal::CommandEncoder::operator->
MTL::ComputeCommandEncoder * operator->()
Definition device.h:61
mlx::core::metal::CommandEncoder::set_input_array
void set_input_array(const array &a, int idx, int64_t offset=0)
mlx::core::metal::CommandEncoder::CommandEncoder
CommandEncoder(const CommandEncoder &)=delete
-
mlx::core::metal::CommandEncoder::outputs
std::unordered_set< const void * > outputs()
Definition device.h:81
-
mlx::core::metal::DeviceStream
Definition device.h:102
-
mlx::core::metal::DeviceStream::~DeviceStream
~DeviceStream()
Definition device.h:104
-
mlx::core::metal::DeviceStream::outputs
std::unordered_map< const void *, std::shared_ptr< Fence > > outputs
Definition device.h:112
-
mlx::core::metal::DeviceStream::DeviceStream
DeviceStream(MTL::CommandQueue *queue)
Definition device.h:103
-
mlx::core::metal::DeviceStream::encoder
std::unique_ptr< CommandEncoder > encoder
Definition device.h:123
-
mlx::core::metal::DeviceStream::fence_mtx
std::mutex fence_mtx
Definition device.h:114
-
mlx::core::metal::DeviceStream::queue
MTL::CommandQueue * queue
Definition device.h:110
-
mlx::core::metal::DeviceStream::fence
std::shared_ptr< Fence > fence
Definition device.h:124
-
mlx::core::metal::DeviceStream::buffer
MTL::CommandBuffer * buffer
Definition device.h:118
-
mlx::core::metal::DeviceStream::buffer_ops
int buffer_ops
Definition device.h:119
-
mlx::core::metal::DeviceStream::temporaries
std::vector< array > temporaries
Definition device.h:125
-
mlx::core::metal::Fence
Definition device.h:94
-
mlx::core::metal::Fence::Fence
Fence(MTL::Fence *fence)
Definition device.h:95
-
mlx::core::metal::Fence::~Fence
~Fence()
Definition device.h:96
-
mlx::core::metal::Fence::fence
MTL::Fence * fence
Definition device.h:99
+ +
mlx::core::metal::CommandEncoder::outputs
std::unordered_set< const void * > outputs()
Definition device.h:82
+
mlx::core::metal::DeviceStream
Definition device.h:105
+
mlx::core::metal::DeviceStream::~DeviceStream
~DeviceStream()
Definition device.h:107
+
mlx::core::metal::DeviceStream::outputs
std::unordered_map< const void *, std::shared_ptr< Fence > > outputs
Definition device.h:115
+
mlx::core::metal::DeviceStream::DeviceStream
DeviceStream(MTL::CommandQueue *queue)
Definition device.h:106
+
mlx::core::metal::DeviceStream::encoder
std::unique_ptr< CommandEncoder > encoder
Definition device.h:126
+
mlx::core::metal::DeviceStream::fence_mtx
std::mutex fence_mtx
Definition device.h:117
+
mlx::core::metal::DeviceStream::queue
MTL::CommandQueue * queue
Definition device.h:113
+
mlx::core::metal::DeviceStream::fence
std::shared_ptr< Fence > fence
Definition device.h:127
+
mlx::core::metal::DeviceStream::buffer
MTL::CommandBuffer * buffer
Definition device.h:121
+
mlx::core::metal::DeviceStream::buffer_ops
int buffer_ops
Definition device.h:122
+
mlx::core::metal::DeviceStream::temporaries
std::vector< array > temporaries
Definition device.h:128
+
mlx::core::metal::Fence
Definition device.h:97
+
mlx::core::metal::Fence::Fence
Fence(MTL::Fence *fence)
Definition device.h:98
+
mlx::core::metal::Fence::~Fence
~Fence()
Definition device.h:99
+
mlx::core::metal::Fence::fence
MTL::Fence * fence
Definition device.h:102
diff --git a/docs/build/html/dev/custom_metal_kernels.html b/docs/build/html/dev/custom_metal_kernels.html index 93bfa9007..51d5e1f7b 100644 --- a/docs/build/html/dev/custom_metal_kernels.html +++ b/docs/build/html/dev/custom_metal_kernels.html @@ -864,7 +864,7 @@
+

Note: grid and threadgroup are parameters to the Metal dispatchThreads function. +This means we will launch mx.prod(grid) threads, subdivided into threadgroup size threadgroups. +For optimal performance, each thread group dimension should be less than or equal to the corresponding grid dimension.

Passing verbose=True to mx.fast.metal_kernel.__call__ will print the generated code for debugging purposes.

diff --git a/docs/build/html/doxygen_crawl.html b/docs/build/html/doxygen_crawl.html index d4fc5d84c..0236914c1 100644 --- a/docs/build/html/doxygen_crawl.html +++ b/docs/build/html/doxygen_crawl.html @@ -4330,9 +4330,9 @@ + - @@ -4443,9 +4443,9 @@ + - @@ -4850,9 +4850,11 @@ + + @@ -4900,7 +4902,6 @@ - @@ -5272,7 +5273,6 @@ - @@ -5933,11 +5933,11 @@ + - @@ -5952,6 +5952,7 @@ + @@ -6016,8 +6017,10 @@ + - + + @@ -6051,7 +6054,7 @@ - + @@ -6929,6 +6932,7 @@ + diff --git a/docs/build/html/functions_func_m.html b/docs/build/html/functions_func_m.html index d4f90d50b..b812c59a5 100644 --- a/docs/build/html/functions_func_m.html +++ b/docs/build/html/functions_func_m.html @@ -93,6 +93,7 @@ $(function(){ initResizable(false); });
  • Matmul() : mlx::core::Matmul
  • max() : metal::_numeric_limits_impl< bfloat16_t >
  • Maximum() : mlx::core::Maximum
    • +
  • maybeInsertBarrier() : mlx::core::metal::CommandEncoder
  • merge_partition() : BlockMergeSort< val_t, idx_t, ARG_SORT, BLOCK_THREADS, N_PER_THREAD, CompareOp >, KernelMultiBlockMergeSort< val_t, idx_t, ARG_SORT, BLOCK_THREADS, N_PER_THREAD, CompareOp >
  • merge_step() : BlockMergeSort< val_t, idx_t, ARG_SORT, BLOCK_THREADS, N_PER_THREAD, CompareOp >
  • min() : metal::_numeric_limits_impl< bfloat16_t >
    • diff --git a/docs/build/html/functions_m.html b/docs/build/html/functions_m.html index 550ae601f..3be734fe6 100644 --- a/docs/build/html/functions_m.html +++ b/docs/build/html/functions_m.html @@ -102,6 +102,7 @@ $(function(){ initResizable(false); });
  • max_exponent : metal::_numeric_limits_impl< bfloat16_t >
  • max_exponent10 : metal::_numeric_limits_impl< bfloat16_t >
  • Maximum() : mlx::core::Maximum
    • +
  • maybeInsertBarrier() : mlx::core::metal::CommandEncoder
  • merge_partition() : BlockMergeSort< val_t, idx_t, ARG_SORT, BLOCK_THREADS, N_PER_THREAD, CompareOp >, KernelMultiBlockMergeSort< val_t, idx_t, ARG_SORT, BLOCK_THREADS, N_PER_THREAD, CompareOp >
  • merge_step() : BlockMergeSort< val_t, idx_t, ARG_SORT, BLOCK_THREADS, N_PER_THREAD, CompareOp >
  • Min : mlx::core::distributed::AllReduce, mlx::core::Reduce, mlx::core::Scan, mlx::core::Scatter
    • diff --git a/docs/build/html/globals_c.html b/docs/build/html/globals_c.html index f2a87df5d..a9eed8c2d 100644 --- a/docs/build/html/globals_c.html +++ b/docs/build/html/globals_c.html @@ -92,8 +92,10 @@ $(function(){ initResizable(false); });
  • can_convert_to_bfloat : bf16.h
  • can_convert_to_complex64 : complex.h
  • ceildiv() : utils.h
    • +
  • col_reduce_2pass() : reduce_col.h
    • +
  • col_reduce_longcolumn() : reduce_col.h
  • col_reduce_looped() : reduce_col.h
    • -
  • col_reduce_small() : reduce_col.h
    • +
  • col_reduce_small() : reduce_col.h
  • complex_binop : complex.h
  • complex_binop_helper : complex.h
  • complex_mul() : radix.h
    • diff --git a/docs/build/html/globals_func_c.html b/docs/build/html/globals_func_c.html index 11545a40d..227276f3e 100644 --- a/docs/build/html/globals_func_c.html +++ b/docs/build/html/globals_func_c.html @@ -88,8 +88,10 @@ $(function(){ initResizable(false); });

    - c -

    diff --git a/docs/build/html/globals_func_s.html b/docs/build/html/globals_func_s.html index d8ecab0f0..ce63bfa00 100644 --- a/docs/build/html/globals_func_s.html +++ b/docs/build/html/globals_func_s.html @@ -88,7 +88,7 @@ $(function(){ initResizable(false); });

    - s -

    diff --git a/docs/build/html/globals_s.html b/docs/build/html/globals_s.html index a46a742a1..00ccc52f6 100644 --- a/docs/build/html/globals_s.html +++ b/docs/build/html/globals_s.html @@ -89,7 +89,7 @@ $(function(){ initResizable(false); });

    - s -

    • scatter_impl() : scatter.h
    • scatter_kernels : indexing.h
      • -
    • sdpa_vector() : sdpa_vector.h
      • +
    • sdpa_vector() : sdpa_vector.h
    • simd_shuffle() : utils.h
    • simd_shuffle_and_fill_up() : utils.h
    • simd_shuffle_down() : utils.h
      • diff --git a/docs/build/html/kernels_8h.html b/docs/build/html/kernels_8h.html index 3f1bb1307..4a5c35d1c 100644 --- a/docs/build/html/kernels_8h.html +++ b/docs/build/html/kernels_8h.html @@ -129,8 +129,8 @@ Functions   MTL::ComputePipelineState * mlx::core::get_mb_sort_kernel (metal::Device &d, const std::string &kernel_name, const array &in, const array &idx, int bn, int tn)   -MTL::ComputePipelineState * mlx::core::get_reduce_init_kernel (metal::Device &d, const std::string &kernel_name, const array &out) -  +MTL::ComputePipelineState * mlx::core::get_reduce_init_kernel (metal::Device &d, const std::string &kernel_name, const std::string &func_name, const std::string &op_name, const array &out) +  MTL::ComputePipelineState * mlx::core::get_reduce_kernel (metal::Device &d, const std::string &kernel_name, const std::string &func_name, const std::string &op_name, const array &in, const array &out, int ndim=-1, int bm=-1, int bn=-1)   MTL::ComputePipelineState * mlx::core::get_steel_gemm_fused_kernel (metal::Device &d, const std::string &kernel_name, const std::string &hash_name, const metal::MTLFCList &func_consts, const array &out, bool transpose_a, bool transpose_b, int bm, int bn, int bk, int wm, int wn) diff --git a/docs/build/html/kernels_8h_source.html b/docs/build/html/kernels_8h_source.html index 93d7a928c..de9608168 100644 --- a/docs/build/html/kernels_8h_source.html +++ b/docs/build/html/kernels_8h_source.html @@ -169,152 +169,154 @@ $(function(){ initResizable(false); });
      76 int bn,
      77 int tn);
      78
      -
      79MTL::ComputePipelineState* get_reduce_init_kernel(
      +
      79MTL::ComputePipelineState* get_reduce_init_kernel(
      80 metal::Device& d,
      81 const std::string& kernel_name,
      -
      82 const array& out);
      -
      83
      -
      84MTL::ComputePipelineState* get_reduce_kernel(
      -
      85 metal::Device& d,
      -
      86 const std::string& kernel_name,
      -
      87 const std::string& func_name,
      -
      88 const std::string& op_name,
      -
      89 const array& in,
      -
      90 const array& out,
      -
      91 int ndim = -1,
      -
      92 int bm = -1,
      -
      93 int bn = -1);
      -
      94
      -
      95MTL::ComputePipelineState* get_steel_gemm_fused_kernel(
      -
      96 metal::Device& d,
      -
      97 const std::string& kernel_name,
      -
      98 const std::string& hash_name,
      -
      99 const metal::MTLFCList& func_consts,
      -
      100 const array& out,
      -
      101 bool transpose_a,
      -
      102 bool transpose_b,
      -
      103 int bm,
      -
      104 int bn,
      -
      105 int bk,
      -
      106 int wm,
      -
      107 int wn);
      -
      108
      -
      109MTL::ComputePipelineState* get_steel_gemm_splitk_kernel(
      -
      110 metal::Device& d,
      -
      111 const std::string& kernel_name,
      -
      112 const array& in,
      -
      113 const array& out,
      -
      114 bool transpose_a,
      -
      115 bool transpose_b,
      -
      116 int bm,
      -
      117 int bn,
      -
      118 int bk,
      -
      119 int wm,
      -
      120 int wn,
      -
      121 bool mn_aligned,
      -
      122 bool k_aligned);
      -
      123
      -
      124MTL::ComputePipelineState* get_steel_gemm_splitk_accum_kernel(
      -
      125 metal::Device& d,
      -
      126 const std::string& kernel_name,
      -
      127 const array& in,
      -
      128 const array& out,
      -
      129 bool axbpy);
      -
      130
      -
      131MTL::ComputePipelineState* get_steel_gemm_masked_kernel(
      -
      132 metal::Device& d,
      -
      133 const std::string& kernel_name,
      -
      134 const array& out,
      -
      135 const std::optional<array>& mask_out,
      -
      136 const std::optional<array>& mask_op,
      -
      137 bool transpose_a,
      -
      138 bool transpose_b,
      -
      139 int bm,
      -
      140 int bn,
      -
      141 int bk,
      -
      142 int wm,
      -
      143 int wn,
      -
      144 bool mn_aligned,
      -
      145 bool k_aligned);
      -
      146
      -
      147MTL::ComputePipelineState* get_steel_conv_kernel(
      -
      148 metal::Device& d,
      -
      149 const std::string& kernel_name,
      -
      150 const array& out,
      -
      151 int bm,
      -
      152 int bn,
      -
      153 int bk,
      -
      154 int wm,
      -
      155 int wn,
      -
      156 int n_channel_specialization,
      -
      157 bool small_filter);
      -
      158
      -
      159MTL::ComputePipelineState* get_gemv_masked_kernel(
      -
      160 metal::Device& d,
      -
      161 const std::string& kernel_name,
      -
      162 const array& out,
      -
      163 const std::optional<array>& mask_out,
      -
      164 const std::optional<array>& mask_op,
      -
      165 bool transpose_mat,
      -
      166 int bm,
      -
      167 int bn,
      -
      168 int sm,
      -
      169 int sn,
      -
      170 int tm,
      -
      171 int tn,
      -
      172 bool contiguous);
      -
      173
      -
      174MTL::ComputePipelineState* get_steel_conv_general_kernel(
      -
      175 metal::Device& d,
      -
      176 const std::string& kernel_name,
      -
      177 const array& out,
      -
      178 int bm,
      -
      179 int bn,
      -
      180 int bk,
      -
      181 int wm,
      -
      182 int wn);
      -
      183
      -
      184MTL::ComputePipelineState* get_fft_kernel(
      -
      185 metal::Device& d,
      -
      186 const std::string& kernel_name,
      -
      187 const std::string& hash_name,
      -
      188 const metal::MTLFCList& func_consts,
      -
      189 const std::string& template_def);
      -
      190
      -
      191MTL::ComputePipelineState* get_quantized_kernel(
      -
      192 metal::Device& d,
      -
      193 const std::string& kernel_name,
      -
      194 const std::string& template_def);
      -
      195
      -
      196// Create a GPU kernel template definition for JIT compilation
      -
      197template <typename... Args>
      -
      198std::string
      -
      -
      199get_template_definition(std::string name, std::string func, Args... args) {
      -
      200 std::ostringstream s;
      -
      201 s << func << "<";
      -
      202 bool first = true;
      -
      203 auto add_arg = [&s, &first](const auto& arg) {
      -
      204 if (!first) {
      -
      205 s << ", ";
      -
      206 }
      -
      207 first = false;
      -
      208 s << arg;
      -
      209 };
      -
      210 (add_arg(args), ...);
      -
      211 s << ">";
      -
      212 return fmt::format(
      -
      213 "\ntemplate [[host_name(\"{0}\")]] [[kernel]] decltype({1}) {1};\n",
      -
      214 name,
      -
      215 s.str());
      -
      216}
      +
      82 const std::string& func_name,
      +
      83 const std::string& op_name,
      +
      84 const array& out);
      +
      85
      +
      86MTL::ComputePipelineState* get_reduce_kernel(
      +
      87 metal::Device& d,
      +
      88 const std::string& kernel_name,
      +
      89 const std::string& func_name,
      +
      90 const std::string& op_name,
      +
      91 const array& in,
      +
      92 const array& out,
      +
      93 int ndim = -1,
      +
      94 int bm = -1,
      +
      95 int bn = -1);
      +
      96
      +
      97MTL::ComputePipelineState* get_steel_gemm_fused_kernel(
      +
      98 metal::Device& d,
      +
      99 const std::string& kernel_name,
      +
      100 const std::string& hash_name,
      +
      101 const metal::MTLFCList& func_consts,
      +
      102 const array& out,
      +
      103 bool transpose_a,
      +
      104 bool transpose_b,
      +
      105 int bm,
      +
      106 int bn,
      +
      107 int bk,
      +
      108 int wm,
      +
      109 int wn);
      +
      110
      +
      111MTL::ComputePipelineState* get_steel_gemm_splitk_kernel(
      +
      112 metal::Device& d,
      +
      113 const std::string& kernel_name,
      +
      114 const array& in,
      +
      115 const array& out,
      +
      116 bool transpose_a,
      +
      117 bool transpose_b,
      +
      118 int bm,
      +
      119 int bn,
      +
      120 int bk,
      +
      121 int wm,
      +
      122 int wn,
      +
      123 bool mn_aligned,
      +
      124 bool k_aligned);
      +
      125
      +
      126MTL::ComputePipelineState* get_steel_gemm_splitk_accum_kernel(
      +
      127 metal::Device& d,
      +
      128 const std::string& kernel_name,
      +
      129 const array& in,
      +
      130 const array& out,
      +
      131 bool axbpy);
      +
      132
      +
      133MTL::ComputePipelineState* get_steel_gemm_masked_kernel(
      +
      134 metal::Device& d,
      +
      135 const std::string& kernel_name,
      +
      136 const array& out,
      +
      137 const std::optional<array>& mask_out,
      +
      138 const std::optional<array>& mask_op,
      +
      139 bool transpose_a,
      +
      140 bool transpose_b,
      +
      141 int bm,
      +
      142 int bn,
      +
      143 int bk,
      +
      144 int wm,
      +
      145 int wn,
      +
      146 bool mn_aligned,
      +
      147 bool k_aligned);
      +
      148
      +
      149MTL::ComputePipelineState* get_steel_conv_kernel(
      +
      150 metal::Device& d,
      +
      151 const std::string& kernel_name,
      +
      152 const array& out,
      +
      153 int bm,
      +
      154 int bn,
      +
      155 int bk,
      +
      156 int wm,
      +
      157 int wn,
      +
      158 int n_channel_specialization,
      +
      159 bool small_filter);
      +
      160
      +
      161MTL::ComputePipelineState* get_gemv_masked_kernel(
      +
      162 metal::Device& d,
      +
      163 const std::string& kernel_name,
      +
      164 const array& out,
      +
      165 const std::optional<array>& mask_out,
      +
      166 const std::optional<array>& mask_op,
      +
      167 bool transpose_mat,
      +
      168 int bm,
      +
      169 int bn,
      +
      170 int sm,
      +
      171 int sn,
      +
      172 int tm,
      +
      173 int tn,
      +
      174 bool contiguous);
      +
      175
      +
      176MTL::ComputePipelineState* get_steel_conv_general_kernel(
      +
      177 metal::Device& d,
      +
      178 const std::string& kernel_name,
      +
      179 const array& out,
      +
      180 int bm,
      +
      181 int bn,
      +
      182 int bk,
      +
      183 int wm,
      +
      184 int wn);
      +
      185
      +
      186MTL::ComputePipelineState* get_fft_kernel(
      +
      187 metal::Device& d,
      +
      188 const std::string& kernel_name,
      +
      189 const std::string& hash_name,
      +
      190 const metal::MTLFCList& func_consts,
      +
      191 const std::string& template_def);
      +
      192
      +
      193MTL::ComputePipelineState* get_quantized_kernel(
      +
      194 metal::Device& d,
      +
      195 const std::string& kernel_name,
      +
      196 const std::string& template_def);
      +
      197
      +
      198// Create a GPU kernel template definition for JIT compilation
      +
      199template <typename... Args>
      +
      200std::string
      +
      +
      201get_template_definition(std::string name, std::string func, Args... args) {
      +
      202 std::ostringstream s;
      +
      203 s << func << "<";
      +
      204 bool first = true;
      +
      205 auto add_arg = [&s, &first](const auto& arg) {
      +
      206 if (!first) {
      +
      207 s << ", ";
      +
      208 }
      +
      209 first = false;
      +
      210 s << arg;
      +
      211 };
      +
      212 (add_arg(args), ...);
      +
      213 s << ">";
      +
      214 return fmt::format(
      +
      215 "\ntemplate [[host_name(\"{0}\")]] [[kernel]] decltype({1}) {1};\n",
      +
      216 name,
      +
      217 s.str());
      +
      218}
      -
      217
      -
      218} // namespace mlx::core
      +
      219
      +
      220} // namespace mlx::core
      mlx::core::array
      Definition array.h:20
      -
      mlx::core::metal::Device
      Definition device.h:128
      +
      mlx::core::metal::Device
      Definition device.h:131
      op
      Op op
      Definition binary.h:129
      mlx::core::metal::MTLFCList
      std::vector< std::tuple< const void *, MTL::DataType, NS::UInteger > > MTLFCList
      Definition device.h:38
      mlx::core
      Definition allocator.h:7
      @@ -322,9 +324,9 @@ $(function(){ initResizable(false); });
      mlx::core::get_steel_gemm_splitk_accum_kernel
      MTL::ComputePipelineState * get_steel_gemm_splitk_accum_kernel(metal::Device &d, const std::string &kernel_name, const array &in, const array &out, bool axbpy)
      mlx::core::get_fft_kernel
      MTL::ComputePipelineState * get_fft_kernel(metal::Device &d, const std::string &kernel_name, const std::string &hash_name, const metal::MTLFCList &func_consts, const std::string &template_def)
      mlx::core::get_softmax_kernel
      MTL::ComputePipelineState * get_softmax_kernel(metal::Device &d, const std::string &kernel_name, bool precise, const array &out)
      +
      mlx::core::get_reduce_init_kernel
      MTL::ComputePipelineState * get_reduce_init_kernel(metal::Device &d, const std::string &kernel_name, const std::string &func_name, const std::string &op_name, const array &out)
      mlx::core::get_binary_kernel
      MTL::ComputePipelineState * get_binary_kernel(metal::Device &d, const std::string &kernel_name, Dtype in_type, Dtype out_type, const std::string op)
      mlx::core::get_binary_two_kernel
      MTL::ComputePipelineState * get_binary_two_kernel(metal::Device &d, const std::string &kernel_name, Dtype in_type, Dtype out_type, const std::string op)
      -
      mlx::core::get_reduce_init_kernel
      MTL::ComputePipelineState * get_reduce_init_kernel(metal::Device &d, const std::string &kernel_name, const array &out)
      mlx::core::get_ternary_kernel
      MTL::ComputePipelineState * get_ternary_kernel(metal::Device &d, const std::string &kernel_name, Dtype type, const std::string op)
      mlx::core::get_arange_kernel
      MTL::ComputePipelineState * get_arange_kernel(metal::Device &d, const std::string &kernel_name, const array &out)
      mlx::core::get_reduce_kernel
      MTL::ComputePipelineState * get_reduce_kernel(metal::Device &d, const std::string &kernel_name, const std::string &func_name, const std::string &op_name, const array &in, const array &out, int ndim=-1, int bm=-1, int bn=-1)
      @@ -332,7 +334,7 @@ $(function(){ initResizable(false); });
      mlx::core::get_steel_gemm_fused_kernel
      MTL::ComputePipelineState * get_steel_gemm_fused_kernel(metal::Device &d, const std::string &kernel_name, const std::string &hash_name, const metal::MTLFCList &func_consts, const array &out, bool transpose_a, bool transpose_b, int bm, int bn, int bk, int wm, int wn)
      mlx::core::get_gemv_masked_kernel
      MTL::ComputePipelineState * get_gemv_masked_kernel(metal::Device &d, const std::string &kernel_name, const array &out, const std::optional< array > &mask_out, const std::optional< array > &mask_op, bool transpose_mat, int bm, int bn, int sm, int sn, int tm, int tn, bool contiguous)
      mlx::core::get_quantized_kernel
      MTL::ComputePipelineState * get_quantized_kernel(metal::Device &d, const std::string &kernel_name, const std::string &template_def)
      -
      mlx::core::get_template_definition
      std::string get_template_definition(std::string name, std::string func, Args... args)
      Definition kernels.h:199
      +
      mlx::core::get_template_definition
      std::string get_template_definition(std::string name, std::string func, Args... args)
      Definition kernels.h:201
      mlx::core::get_steel_gemm_masked_kernel
      MTL::ComputePipelineState * get_steel_gemm_masked_kernel(metal::Device &d, const std::string &kernel_name, const array &out, const std::optional< array > &mask_out, const std::optional< array > &mask_op, bool transpose_a, bool transpose_b, int bm, int bn, int bk, int wm, int wn, bool mn_aligned, bool k_aligned)
      mlx::core::get_steel_conv_general_kernel
      MTL::ComputePipelineState * get_steel_conv_general_kernel(metal::Device &d, const std::string &kernel_name, const array &out, int bm, int bn, int bk, int wm, int wn)
      mlx::core::get_steel_conv_kernel
      MTL::ComputePipelineState * get_steel_conv_kernel(metal::Device &d, const std::string &kernel_name, const array &out, int bm, int bn, int bk, int wm, int wn, int n_channel_specialization, bool small_filter)
      diff --git a/docs/build/html/matmul_8h_source.html b/docs/build/html/matmul_8h_source.html index 4258466e4..ce4ccfe53 100644 --- a/docs/build/html/matmul_8h_source.html +++ b/docs/build/html/matmul_8h_source.html @@ -143,7 +143,7 @@ $(function(){ initResizable(false); });
      50} // namespace mlx::core
      mlx::core::array
      Definition array.h:20
      -
      mlx::core::metal::Device
      Definition device.h:128
      +
      mlx::core::metal::Device
      Definition device.h:131
      mlx::core
      Definition allocator.h:7
      mlx::core::steel_matmul_regular
      void steel_matmul_regular(const Stream &s, metal::Device &d, const array &a, const array &b, array &out, int M, int N, int K, int batch_size_out, int lda, int ldb, int ldd, bool transpose_a, bool transpose_b, std::vector< int > batch_shape, std::vector< size_t > batch_strides, size_t A_batch_stride, size_t B_batch_stride, size_t matrix_stride_out, std::vector< array > &copies)
      mlx::core::steel_matmul
      void steel_matmul(const Stream &s, metal::Device &d, const array &a, const array &b, array &out, int M, int N, int K, int batch_size_out, int lda, int ldb, bool transpose_a, bool transpose_b, std::vector< array > &copies, std::vector< int > batch_shape={}, std::vector< size_t > A_batch_stride={}, std::vector< size_t > B_batch_stride={})
      diff --git a/docs/build/html/metal_2reduce_8h.html b/docs/build/html/metal_2reduce_8h.html index 43294f485..d08ed5e85 100644 --- a/docs/build/html/metal_2reduce_8h.html +++ b/docs/build/html/metal_2reduce_8h.html @@ -109,8 +109,8 @@ Namespaces - - + + diff --git a/docs/build/html/metal_2reduce_8h_source.html b/docs/build/html/metal_2reduce_8h_source.html index 9dfc6469d..79e30d6a2 100644 --- a/docs/build/html/metal_2reduce_8h_source.html +++ b/docs/build/html/metal_2reduce_8h_source.html @@ -103,44 +103,43 @@ $(function(){ initResizable(false); });
      10
      11using metal::CommandEncoder;
      12
      -
      13void all_reduce_dispatch(
      +
      13void all_reduce_dispatch(
      14 const array& in,
      15 array& out,
      16 const std::string& op_name,
      17 CommandEncoder& compute_encoder,
      18 metal::Device& d,
      -
      19 const Stream& s,
      -
      20 std::vector<array>& copies);
      -
      21
      -
      22void row_reduce_general_dispatch(
      -
      23 const array& in,
      -
      24 array& out,
      -
      25 const std::string& op_name,
      -
      26 const ReductionPlan& plan,
      -
      27 const std::vector<int>& axes,
      -
      28 CommandEncoder& compute_encoder,
      -
      29 metal::Device& d,
      -
      30 const Stream& s);
      -
      31
      -
      32void strided_reduce_general_dispatch(
      -
      33 const array& in,
      -
      34 array& out,
      -
      35 const std::string& op_name,
      -
      36 const ReductionPlan& plan,
      -
      37 const std::vector<int>& axes,
      -
      38 CommandEncoder& compute_encoder,
      -
      39 metal::Device& d,
      -
      40 const Stream& s);
      -
      41
      -
      42} // namespace mlx::core
      +
      19 const Stream& s);
      +
      20
      +
      21void row_reduce_general_dispatch(
      +
      22 const array& in,
      +
      23 array& out,
      +
      24 const std::string& op_name,
      +
      25 const ReductionPlan& plan,
      +
      26 const std::vector<int>& axes,
      +
      27 CommandEncoder& compute_encoder,
      +
      28 metal::Device& d,
      +
      29 const Stream& s);
      +
      30
      +
      31void strided_reduce_general_dispatch(
      +
      32 const array& in,
      +
      33 array& out,
      +
      34 const std::string& op_name,
      +
      35 const ReductionPlan& plan,
      +
      36 const std::vector<int>& axes,
      +
      37 CommandEncoder& compute_encoder,
      +
      38 metal::Device& d,
      +
      39 const Stream& s);
      +
      40
      +
      41} // namespace mlx::core
      mlx::core::array
      Definition array.h:20
      -
      mlx::core::metal::Device
      Definition device.h:128
      +
      mlx::core::metal::Device
      Definition device.h:131
      mlx::core
      Definition allocator.h:7
      +
      mlx::core::all_reduce_dispatch
      void all_reduce_dispatch(const array &in, array &out, const std::string &op_name, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s)
      mlx::core::strided_reduce_general_dispatch
      void strided_reduce_general_dispatch(const array &in, array &out, const std::string &op_name, const ReductionPlan &plan, const std::vector< int > &axes, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s)
      mlx::core::row_reduce_general_dispatch
      void row_reduce_general_dispatch(const array &in, array &out, const std::string &op_name, const ReductionPlan &plan, const std::vector< int > &axes, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s)
      -
      mlx::core::all_reduce_dispatch
      void all_reduce_dispatch(const array &in, array &out, const std::string &op_name, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s, std::vector< array > &copies)
      mlx::core::ReductionPlan
      Definition reduce.h:39
      mlx::core::Stream
      Definition stream.h:9
      diff --git a/docs/build/html/namespacemembers.html b/docs/build/html/namespacemembers.html index ffff1cc90..5625555af 100644 --- a/docs/build/html/namespacemembers.html +++ b/docs/build/html/namespacemembers.html @@ -99,7 +99,7 @@ $(function(){ initResizable(false); });
    • aligned_dealloc() : pocketfft::detail
    • all() : mlx::core
    • all_gather() : mlx::core::distributed, mlx::core::distributed::detail
      • -
    • all_reduce_dispatch() : mlx::core
      • +
    • all_reduce_dispatch() : mlx::core
    • all_sum() : mlx::core::distributed, mlx::core::distributed::detail
    • allclose() : mlx::core
    • alloc_tmp() : pocketfft::detail
      • diff --git a/docs/build/html/namespacemembers_func.html b/docs/build/html/namespacemembers_func.html index 9e16a9a69..e838b45c1 100644 --- a/docs/build/html/namespacemembers_func.html +++ b/docs/build/html/namespacemembers_func.html @@ -98,7 +98,7 @@ $(function(){ initResizable(false); });
    • aligned_dealloc() : pocketfft::detail
    • all() : mlx::core
    • all_gather() : mlx::core::distributed, mlx::core::distributed::detail
      • -
    • all_reduce_dispatch() : mlx::core
      • +
    • all_reduce_dispatch() : mlx::core
    • all_sum() : mlx::core::distributed, mlx::core::distributed::detail
    • allclose() : mlx::core
    • alloc_tmp() : pocketfft::detail
      • diff --git a/docs/build/html/namespacemembers_func_g.html b/docs/build/html/namespacemembers_func_g.html index 5e2c379ff..1c777baeb 100644 --- a/docs/build/html/namespacemembers_func_g.html +++ b/docs/build/html/namespacemembers_func_g.html @@ -112,7 +112,7 @@ $(function(){ initResizable(false); });
    • get_pool() : pocketfft::detail::threading
    • get_primitive_string() : mlx::core
    • get_quantized_kernel() : mlx::core
      • -
    • get_reduce_init_kernel() : mlx::core
      • +
    • get_reduce_init_kernel() : mlx::core
    • get_reduce_kernel() : mlx::core
    • get_reduction_plan() : mlx::core
    • get_scan_kernel() : mlx::core
      • diff --git a/docs/build/html/namespacemembers_g.html b/docs/build/html/namespacemembers_g.html index 73be65412..0bbdf9db4 100644 --- a/docs/build/html/namespacemembers_g.html +++ b/docs/build/html/namespacemembers_g.html @@ -116,7 +116,7 @@ $(function(){ initResizable(false); });
    • get_pool() : pocketfft::detail::threading
    • get_primitive_string() : mlx::core
    • get_quantized_kernel() : mlx::core
      • -
    • get_reduce_init_kernel() : mlx::core
      • +
    • get_reduce_init_kernel() : mlx::core
    • get_reduce_kernel() : mlx::core
    • get_reduction_plan() : mlx::core
    • get_scan_kernel() : mlx::core
      • diff --git a/docs/build/html/namespacemlx_1_1core.html b/docs/build/html/namespacemlx_1_1core.html index cef5d8c69..09ca616f9 100644 --- a/docs/build/html/namespacemlx_1_1core.html +++ b/docs/build/html/namespacemlx_1_1core.html @@ -534,8 +534,8 @@ Functions - - + + @@ -563,8 +563,8 @@ Functions - - + + @@ -2634,8 +2634,8 @@ template<typename... T>

      Function Documentation

      - -

      ◆ all_reduce_dispatch()

      + +

      ◆ all_reduce_dispatch()

      @@ -2668,12 +2668,7 @@ template<typename... T>
      - - - - - - +

      Functions

      void mlx::core::all_reduce_dispatch (const array &in, array &out, const std::string &op_name, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s, std::vector< array > &copies)
       
      void mlx::core::all_reduce_dispatch (const array &in, array &out, const std::string &op_name, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s)
       
      void mlx::core::row_reduce_general_dispatch (const array &in, array &out, const std::string &op_name, const ReductionPlan &plan, const std::vector< int > &axes, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s)
       
      void mlx::core::strided_reduce_general_dispatch (const array &in, array &out, const std::string &op_name, const ReductionPlan &plan, const std::vector< int > &axes, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s)
       
      MTL::ComputePipelineState * get_mb_sort_kernel (metal::Device &d, const std::string &kernel_name, const array &in, const array &idx, int bn, int tn)
       
      MTL::ComputePipelineState * get_reduce_init_kernel (metal::Device &d, const std::string &kernel_name, const array &out)
       
      MTL::ComputePipelineState * get_reduce_init_kernel (metal::Device &d, const std::string &kernel_name, const std::string &func_name, const std::string &op_name, const array &out)
       
      MTL::ComputePipelineState * get_reduce_kernel (metal::Device &d, const std::string &kernel_name, const std::string &func_name, const std::string &op_name, const array &in, const array &out, int ndim=-1, int bm=-1, int bn=-1)
       
      MTL::ComputePipelineState * get_steel_gemm_fused_kernel (metal::Device &d, const std::string &kernel_name, const std::string &hash_name, const metal::MTLFCList &func_consts, const array &out, bool transpose_a, bool transpose_b, int bm, int bn, int bk, int wm, int wn)
       
      void steel_matmul (const Stream &s, metal::Device &d, const array &a, const array &b, array &out, int M, int N, int K, int batch_size_out, int lda, int ldb, bool transpose_a, bool transpose_b, std::vector< array > &copies, std::vector< int > batch_shape={}, std::vector< size_t > A_batch_stride={}, std::vector< size_t > B_batch_stride={})
       
      void all_reduce_dispatch (const array &in, array &out, const std::string &op_name, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s, std::vector< array > &copies)
       
      void all_reduce_dispatch (const array &in, array &out, const std::string &op_name, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s)
       
      void row_reduce_general_dispatch (const array &in, array &out, const std::string &op_name, const ReductionPlan &plan, const std::vector< int > &axes, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s)
       
      void strided_reduce_general_dispatch (const array &in, array &out, const std::string &op_name, const ReductionPlan &plan, const std::vector< int > &axes, CommandEncoder &compute_encoder, metal::Device &d, const Stream &s)
      const Stream & s,
      std::vector< array > & copies )const Stream & s )
      @@ -4418,8 +4413,8 @@ template<typename... Arrays, typename = enable_for_arrays_t<Arrays...>
    - -

    ◆ get_reduce_init_kernel()

    + +

    ◆ get_reduce_init_kernel()

    @@ -4434,6 +4429,16 @@ template<typename... Arrays, typename = enable_for_arrays_t<Arrays...> const std::string & kernel_name, + + + + const std::string & func_name, + + + + + const std::string & op_name, + diff --git a/docs/build/html/objects.inv b/docs/build/html/objects.inv index c31b0569f..5bbbf441f 100644 Binary files a/docs/build/html/objects.inv and b/docs/build/html/objects.inv differ diff --git a/docs/build/html/python/_autosummary/mlx.core.fast.metal_kernel.html b/docs/build/html/python/_autosummary/mlx.core.fast.metal_kernel.html index 085ae4f4a..a69f99808 100644 --- a/docs/build/html/python/_autosummary/mlx.core.fast.metal_kernel.html +++ b/docs/build/html/python/_autosummary/mlx.core.fast.metal_kernel.html @@ -867,6 +867,7 @@
    metal_kernel(name: str, input_names: Sequence[str], output_names: Sequence[str], source: str, header: str = '', ensure_row_contiguous: bool = True, atomic_outputs: bool = False) → object#

    A jit-compiled custom Metal kernel defined from a source string.

    +

    Full documentation: Custom Metal Kernels.

    Parameters:
      diff --git a/docs/build/html/quantized_8h.html b/docs/build/html/quantized_8h.html index f865582b9..c0e18903c 100644 --- a/docs/build/html/quantized_8h.html +++ b/docs/build/html/quantized_8h.html @@ -140,9 +140,9 @@ Functions template<typename T , int group_size, int bits> METAL_FUNC void qmv_impl (const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, uint3 tid, uint simd_gid, uint simd_lid)   -template<typename T , const int group_size, const int bits> -METAL_FUNC void qvm_impl (const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, uint3 tid, uint simd_gid, uint simd_lid) -  +template<typename T , const int group_size, const int bits> +METAL_FUNC void qvm_impl (const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const int in_vec_size, const int out_vec_size, uint3 tid, uint simd_gid, uint simd_lid) +  template<typename T , const int group_size, const int bits, const bool aligned_N, const int BM = 32, const int BK = 32, const int BN = 32> METAL_FUNC void qmm_t_impl (const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, threadgroup T *Xs, threadgroup T *Ws, const constant int &K, const constant int &N, const constant int &M, uint3 tid, uint lid, uint simd_gid, uint simd_lid)   @@ -167,6 +167,9 @@ Functions template<typename T , const int group_size, const int bits, bool batched> void qvm (const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, uint3 tid, uint simd_gid, uint simd_lid)   +template<typename T , const int group_size, const int bits, int split_k = 32> +void qvm_split_k (const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &final_block_size, uint3 tid, uint simd_gid, uint simd_lid) +  template<typename T , const int group_size, const int bits, const bool aligned_N, const bool batched, const int BM = 32, const int BK = 32, const int BN = 32> void qmm_t (const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &K, const constant int &N, const constant int &M, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, uint3 tid, uint lid, uint simd_gid, uint simd_lid)   @@ -2485,8 +2488,8 @@ template<typename T , const int group_size, const int bits, bool batched>
    - -

    ◆ qvm_impl()

    + +

    ◆ qvm_impl()

    @@ -2518,6 +2521,69 @@ template<typename T , const int group_size, const int bits>
    device T * y, + + + + const int in_vec_size, + + + + + const int out_vec_size, + + + + + uint3 tid, + + + + + uint simd_gid, + + + + + uint simd_lid ) + + +
    + +
    +
    + +

    ◆ qvm_split_k()

    + +
    +
    +
    +template<typename T , const int group_size, const int bits, int split_k = 32>
    + + + + + + + + + + + + + + + + + + + + + + + + + + @@ -2528,6 +2594,51 @@ template<typename T , const int group_size, const int bits> + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/docs/build/html/quantized_8h_source.html b/docs/build/html/quantized_8h_source.html index 07b25d33f..0a08db757 100644 --- a/docs/build/html/quantized_8h_source.html +++ b/docs/build/html/quantized_8h_source.html @@ -766,14 +766,14 @@ $(function(){ initResizable(false); });
    645
    646template <typename T, const int group_size, const int bits>
    -
    647METAL_FUNC void qvm_impl(
    +
    647METAL_FUNC void qvm_impl(
    648 const device uint32_t* w,
    649 const device T* scales,
    650 const device T* biases,
    651 const device T* x,
    652 device T* y,
    -
    653 const constant int& in_vec_size,
    -
    654 const constant int& out_vec_size,
    +
    653 const int in_vec_size,
    +
    654 const int out_vec_size,
    655 uint3 tid [[threadgroup_position_in_grid]],
    656 uint simd_gid [[simdgroup_index_in_threadgroup]],
    657 uint simd_lid [[thread_index_in_simdgroup]]) {
    @@ -1423,7 +1423,7 @@ $(function(){ initResizable(false); });
    1285 b_strides,
    1286 tid);
    1287 }
    -
    1288 qvm_impl<T, group_size, bits>(
    +
    1288 qvm_impl<T, group_size, bits>(
    1289 w,
    1290 scales,
    1291 biases,
    @@ -1437,610 +1437,667 @@ $(function(){ initResizable(false); });
    1299}
    1300
    -
    1301template <
    -
    1302 typename T,
    -
    1303 const int group_size,
    -
    1304 const int bits,
    -
    1305 const bool aligned_N,
    -
    1306 const bool batched,
    -
    1307 const int BM = 32,
    -
    1308 const int BK = 32,
    -
    1309 const int BN = 32>
    -
    -
    1310[[kernel]] void qmm_t(
    -
    1311 const device uint32_t* w [[buffer(0)]],
    -
    1312 const device T* scales [[buffer(1)]],
    -
    1313 const device T* biases [[buffer(2)]],
    -
    1314 const device T* x [[buffer(3)]],
    -
    1315 device T* y [[buffer(4)]],
    -
    1316 const constant int& K [[buffer(5)]],
    -
    1317 const constant int& N [[buffer(6)]],
    -
    1318 const constant int& M [[buffer(7)]],
    -
    1319 const constant int& x_batch_ndims [[buffer(8)]],
    -
    1320 const constant int* x_shape [[buffer(9)]],
    -
    1321 const constant size_t* x_strides [[buffer(10)]],
    -
    1322 const constant int& w_batch_ndims [[buffer(11)]],
    -
    1323 const constant int* w_shape [[buffer(12)]],
    -
    1324 const constant size_t* w_strides [[buffer(13)]],
    -
    1325 const constant size_t* s_strides [[buffer(14)]],
    -
    1326 const constant size_t* b_strides [[buffer(15)]],
    -
    1327 uint3 tid [[threadgroup_position_in_grid]],
    -
    1328 uint lid [[thread_index_in_threadgroup]],
    -
    1329 uint simd_gid [[simdgroup_index_in_threadgroup]],
    -
    1330 uint simd_lid [[thread_index_in_simdgroup]]) {
    -
    1331 (void)lid;
    -
    1332
    -
    1333 constexpr int BK_padded = (BK + 16 / sizeof(T));
    -
    1334
    -
    1335 threadgroup T Xs[BM * BK_padded];
    -
    1336 threadgroup T Ws[BN * BK_padded];
    -
    1337
    -
    1338 if (batched) {
    -
    1339 adjust_matrix_offsets<T>(
    -
    1340 x,
    -
    1341 w,
    -
    1342 scales,
    -
    1343 biases,
    -
    1344 y,
    -
    1345 M * N,
    -
    1346 x_batch_ndims,
    -
    1347 x_shape,
    -
    1348 x_strides,
    -
    1349 w_batch_ndims,
    -
    1350 w_shape,
    -
    1351 w_strides,
    -
    1352 s_strides,
    -
    1353 b_strides,
    -
    1354 tid);
    -
    1355 }
    -
    1356 qmm_t_impl<T, group_size, bits, aligned_N, BM, BK, BN>(
    -
    1357 w, scales, biases, x, y, Xs, Ws, K, N, M, tid, lid, simd_gid, simd_lid);
    -
    1358}
    +
    1301template <typename T, const int group_size, const int bits, int split_k = 32>
    +
    +
    1302[[kernel]] void qvm_split_k(
    +
    1303 const device uint32_t* w [[buffer(0)]],
    +
    1304 const device T* scales [[buffer(1)]],
    +
    1305 const device T* biases [[buffer(2)]],
    +
    1306 const device T* x [[buffer(3)]],
    +
    1307 device T* y [[buffer(4)]],
    +
    1308 const constant int& in_vec_size [[buffer(5)]],
    +
    1309 const constant int& out_vec_size [[buffer(6)]],
    +
    1310 const constant int& x_batch_ndims [[buffer(7)]],
    +
    1311 const constant int* x_shape [[buffer(8)]],
    +
    1312 const constant size_t* x_strides [[buffer(9)]],
    +
    1313 const constant int& w_batch_ndims [[buffer(10)]],
    +
    1314 const constant int* w_shape [[buffer(11)]],
    +
    1315 const constant size_t* w_strides [[buffer(12)]],
    +
    1316 const constant size_t* s_strides [[buffer(13)]],
    +
    1317 const constant size_t* b_strides [[buffer(14)]],
    +
    1318 const constant int& final_block_size [[buffer(15)]],
    +
    1319 uint3 tid [[threadgroup_position_in_grid]],
    +
    1320 uint simd_gid [[simdgroup_index_in_threadgroup]],
    +
    1321 uint simd_lid [[thread_index_in_simdgroup]]) {
    +
    1322 adjust_matrix_offsets<T>(
    +
    1323 x,
    +
    1324 w,
    +
    1325 scales,
    +
    1326 biases,
    +
    1327 y,
    +
    1328 out_vec_size,
    +
    1329 x_batch_ndims,
    +
    1330 x_shape,
    +
    1331 x_strides,
    +
    1332 w_batch_ndims,
    +
    1333 w_shape,
    +
    1334 w_strides,
    +
    1335 s_strides,
    +
    1336 b_strides,
    +
    1337 tid);
    +
    1338
    +
    1339 // When (in_vec_size % split_k != 0) the final block needs to be smaller
    +
    1340 int in_vec_size_adj =
    +
    1341 tid.z % split_k == split_k - 1 ? final_block_size : in_vec_size;
    +
    1342
    +
    1343 qvm_impl<T, group_size, bits>(
    +
    1344 w,
    +
    1345 scales,
    +
    1346 biases,
    +
    1347 x,
    +
    1348 y,
    +
    1349 in_vec_size_adj,
    +
    1350 out_vec_size,
    +
    1351 tid,
    +
    1352 simd_gid,
    +
    1353 simd_lid);
    +
    1354}
    -
    1359
    -
    1360template <
    -
    1361 typename T,
    -
    1362 const int group_size,
    -
    1363 const int bits,
    -
    1364 const bool batched,
    -
    1365 const int BM = 32,
    -
    1366 const int BK = 32,
    -
    1367 const int BN = 32>
    -
    -
    1368[[kernel]] void qmm_n(
    -
    1369 const device uint32_t* w [[buffer(0)]],
    -
    1370 const device T* scales [[buffer(1)]],
    -
    1371 const device T* biases [[buffer(2)]],
    -
    1372 const device T* x [[buffer(3)]],
    -
    1373 device T* y [[buffer(4)]],
    -
    1374 const constant int& K [[buffer(5)]],
    -
    1375 const constant int& N [[buffer(6)]],
    -
    1376 const constant int& M [[buffer(7)]],
    -
    1377 const constant int& x_batch_ndims [[buffer(8)]],
    -
    1378 const constant int* x_shape [[buffer(9)]],
    -
    1379 const constant size_t* x_strides [[buffer(10)]],
    -
    1380 const constant int& w_batch_ndims [[buffer(11)]],
    -
    1381 const constant int* w_shape [[buffer(12)]],
    -
    1382 const constant size_t* w_strides [[buffer(13)]],
    -
    1383 const constant size_t* s_strides [[buffer(14)]],
    -
    1384 const constant size_t* b_strides [[buffer(15)]],
    -
    1385 uint3 tid [[threadgroup_position_in_grid]],
    -
    1386 uint lid [[thread_index_in_threadgroup]],
    -
    1387 uint simd_gid [[simdgroup_index_in_threadgroup]],
    -
    1388 uint simd_lid [[thread_index_in_simdgroup]]) {
    -
    1389 (void)lid;
    -
    1390
    -
    1391 constexpr int BK_padded = (BK + 16 / sizeof(T));
    -
    1392 constexpr int BN_padded = (BN + 16 / sizeof(T));
    -
    1393
    -
    1394 threadgroup T Xs[BM * BK_padded];
    -
    1395 threadgroup T Ws[BK * BN_padded];
    -
    1396
    -
    1397 if (batched) {
    -
    1398 adjust_matrix_offsets<T>(
    -
    1399 x,
    -
    1400 w,
    -
    1401 scales,
    -
    1402 biases,
    -
    1403 y,
    -
    1404 M * N,
    -
    1405 x_batch_ndims,
    -
    1406 x_shape,
    -
    1407 x_strides,
    -
    1408 w_batch_ndims,
    -
    1409 w_shape,
    -
    1410 w_strides,
    -
    1411 s_strides,
    -
    1412 b_strides,
    -
    1413 tid);
    -
    1414 }
    -
    1415
    -
    1416 qmm_n_impl<T, group_size, bits, BM, BK, BN>(
    -
    1417 w, scales, biases, x, y, Xs, Ws, K, N, M, tid, lid, simd_gid, simd_lid);
    -
    1418}
    +
    1355
    +
    1356template <
    +
    1357 typename T,
    +
    1358 const int group_size,
    +
    1359 const int bits,
    +
    1360 const bool aligned_N,
    +
    1361 const bool batched,
    +
    1362 const int BM = 32,
    +
    1363 const int BK = 32,
    +
    1364 const int BN = 32>
    +
    +
    1365[[kernel]] void qmm_t(
    +
    1366 const device uint32_t* w [[buffer(0)]],
    +
    1367 const device T* scales [[buffer(1)]],
    +
    1368 const device T* biases [[buffer(2)]],
    +
    1369 const device T* x [[buffer(3)]],
    +
    1370 device T* y [[buffer(4)]],
    +
    1371 const constant int& K [[buffer(5)]],
    +
    1372 const constant int& N [[buffer(6)]],
    +
    1373 const constant int& M [[buffer(7)]],
    +
    1374 const constant int& x_batch_ndims [[buffer(8)]],
    +
    1375 const constant int* x_shape [[buffer(9)]],
    +
    1376 const constant size_t* x_strides [[buffer(10)]],
    +
    1377 const constant int& w_batch_ndims [[buffer(11)]],
    +
    1378 const constant int* w_shape [[buffer(12)]],
    +
    1379 const constant size_t* w_strides [[buffer(13)]],
    +
    1380 const constant size_t* s_strides [[buffer(14)]],
    +
    1381 const constant size_t* b_strides [[buffer(15)]],
    +
    1382 uint3 tid [[threadgroup_position_in_grid]],
    +
    1383 uint lid [[thread_index_in_threadgroup]],
    +
    1384 uint simd_gid [[simdgroup_index_in_threadgroup]],
    +
    1385 uint simd_lid [[thread_index_in_simdgroup]]) {
    +
    1386 (void)lid;
    +
    1387
    +
    1388 constexpr int BK_padded = (BK + 16 / sizeof(T));
    +
    1389
    +
    1390 threadgroup T Xs[BM * BK_padded];
    +
    1391 threadgroup T Ws[BN * BK_padded];
    +
    1392
    +
    1393 if (batched) {
    +
    1394 adjust_matrix_offsets<T>(
    +
    1395 x,
    +
    1396 w,
    +
    1397 scales,
    +
    1398 biases,
    +
    1399 y,
    +
    1400 M * N,
    +
    1401 x_batch_ndims,
    +
    1402 x_shape,
    +
    1403 x_strides,
    +
    1404 w_batch_ndims,
    +
    1405 w_shape,
    +
    1406 w_strides,
    +
    1407 s_strides,
    +
    1408 b_strides,
    +
    1409 tid);
    +
    1410 }
    +
    1411 qmm_t_impl<T, group_size, bits, aligned_N, BM, BK, BN>(
    +
    1412 w, scales, biases, x, y, Xs, Ws, K, N, M, tid, lid, simd_gid, simd_lid);
    +
    1413}
    -
    1419
    -
    1420template <typename T, int group_size, int bits>
    -
    -
    1421[[kernel]] void bs_qmv_fast(
    -
    1422 const device uint32_t* w [[buffer(0)]],
    -
    1423 const device T* scales [[buffer(1)]],
    -
    1424 const device T* biases [[buffer(2)]],
    -
    1425 const device T* x [[buffer(3)]],
    -
    1426 device T* y [[buffer(4)]],
    -
    1427 const constant int& in_vec_size [[buffer(5)]],
    -
    1428 const constant int& out_vec_size [[buffer(6)]],
    -
    1429 const constant int& x_batch_ndims [[buffer(7)]],
    -
    1430 const constant int* x_shape [[buffer(8)]],
    -
    1431 const constant size_t* x_strides [[buffer(9)]],
    -
    1432 const constant int& w_batch_ndims [[buffer(10)]],
    -
    1433 const constant int* w_shape [[buffer(11)]],
    -
    1434 const constant size_t* w_strides [[buffer(12)]],
    -
    1435 const constant size_t* s_strides [[buffer(13)]],
    -
    1436 const constant size_t* b_strides [[buffer(14)]],
    -
    1437 const constant int& batch_ndims [[buffer(15)]],
    -
    1438 const constant int* batch_shape [[buffer(16)]],
    -
    1439 const device uint32_t* lhs_indices [[buffer(17)]],
    -
    1440 const device uint32_t* rhs_indices [[buffer(18)]],
    -
    1441 const constant size_t* lhs_strides [[buffer(19)]],
    -
    1442 const constant size_t* rhs_strides [[buffer(20)]],
    -
    1443 uint3 tid [[threadgroup_position_in_grid]],
    -
    1444 uint simd_gid [[simdgroup_index_in_threadgroup]],
    -
    1445 uint simd_lid [[thread_index_in_simdgroup]]) {
    -
    1446 adjust_matrix_offsets<T>(
    -
    1447 x,
    -
    1448 w,
    -
    1449 scales,
    -
    1450 biases,
    -
    1451 lhs_indices,
    -
    1452 rhs_indices,
    -
    1453 y,
    -
    1454 out_vec_size,
    -
    1455 batch_ndims,
    -
    1456 batch_shape,
    -
    1457 lhs_strides,
    -
    1458 rhs_strides,
    -
    1459 x_batch_ndims,
    -
    1460 x_shape,
    -
    1461 x_strides,
    -
    1462 w_batch_ndims,
    -
    1463 w_shape,
    -
    1464 w_strides,
    -
    1465 s_strides,
    -
    1466 b_strides,
    -
    1467 tid);
    -
    1468 qmv_fast_impl<T, group_size, bits>(
    -
    1469 w,
    -
    1470 scales,
    -
    1471 biases,
    -
    1472 x,
    -
    1473 y,
    -
    1474 in_vec_size,
    -
    1475 out_vec_size,
    -
    1476 tid,
    -
    1477 simd_gid,
    -
    1478 simd_lid);
    -
    1479}
    +
    1414
    +
    1415template <
    +
    1416 typename T,
    +
    1417 const int group_size,
    +
    1418 const int bits,
    +
    1419 const bool batched,
    +
    1420 const int BM = 32,
    +
    1421 const int BK = 32,
    +
    1422 const int BN = 32>
    +
    +
    1423[[kernel]] void qmm_n(
    +
    1424 const device uint32_t* w [[buffer(0)]],
    +
    1425 const device T* scales [[buffer(1)]],
    +
    1426 const device T* biases [[buffer(2)]],
    +
    1427 const device T* x [[buffer(3)]],
    +
    1428 device T* y [[buffer(4)]],
    +
    1429 const constant int& K [[buffer(5)]],
    +
    1430 const constant int& N [[buffer(6)]],
    +
    1431 const constant int& M [[buffer(7)]],
    +
    1432 const constant int& x_batch_ndims [[buffer(8)]],
    +
    1433 const constant int* x_shape [[buffer(9)]],
    +
    1434 const constant size_t* x_strides [[buffer(10)]],
    +
    1435 const constant int& w_batch_ndims [[buffer(11)]],
    +
    1436 const constant int* w_shape [[buffer(12)]],
    +
    1437 const constant size_t* w_strides [[buffer(13)]],
    +
    1438 const constant size_t* s_strides [[buffer(14)]],
    +
    1439 const constant size_t* b_strides [[buffer(15)]],
    +
    1440 uint3 tid [[threadgroup_position_in_grid]],
    +
    1441 uint lid [[thread_index_in_threadgroup]],
    +
    1442 uint simd_gid [[simdgroup_index_in_threadgroup]],
    +
    1443 uint simd_lid [[thread_index_in_simdgroup]]) {
    +
    1444 (void)lid;
    +
    1445
    +
    1446 constexpr int BK_padded = (BK + 16 / sizeof(T));
    +
    1447 constexpr int BN_padded = (BN + 16 / sizeof(T));
    +
    1448
    +
    1449 threadgroup T Xs[BM * BK_padded];
    +
    1450 threadgroup T Ws[BK * BN_padded];
    +
    1451
    +
    1452 if (batched) {
    +
    1453 adjust_matrix_offsets<T>(
    +
    1454 x,
    +
    1455 w,
    +
    1456 scales,
    +
    1457 biases,
    +
    1458 y,
    +
    1459 M * N,
    +
    1460 x_batch_ndims,
    +
    1461 x_shape,
    +
    1462 x_strides,
    +
    1463 w_batch_ndims,
    +
    1464 w_shape,
    +
    1465 w_strides,
    +
    1466 s_strides,
    +
    1467 b_strides,
    +
    1468 tid);
    +
    1469 }
    +
    1470
    +
    1471 qmm_n_impl<T, group_size, bits, BM, BK, BN>(
    +
    1472 w, scales, biases, x, y, Xs, Ws, K, N, M, tid, lid, simd_gid, simd_lid);
    +
    1473}
    -
    1480
    -
    1481template <typename T, int group_size, int bits>
    -
    -
    1482[[kernel]] void bs_qmv(
    -
    1483 const device uint32_t* w [[buffer(0)]],
    -
    1484 const device T* scales [[buffer(1)]],
    -
    1485 const device T* biases [[buffer(2)]],
    -
    1486 const device T* x [[buffer(3)]],
    -
    1487 device T* y [[buffer(4)]],
    -
    1488 const constant int& in_vec_size [[buffer(5)]],
    -
    1489 const constant int& out_vec_size [[buffer(6)]],
    -
    1490 const constant int& x_batch_ndims [[buffer(7)]],
    -
    1491 const constant int* x_shape [[buffer(8)]],
    -
    1492 const constant size_t* x_strides [[buffer(9)]],
    -
    1493 const constant int& w_batch_ndims [[buffer(10)]],
    -
    1494 const constant int* w_shape [[buffer(11)]],
    -
    1495 const constant size_t* w_strides [[buffer(12)]],
    -
    1496 const constant size_t* s_strides [[buffer(13)]],
    -
    1497 const constant size_t* b_strides [[buffer(14)]],
    -
    1498 const constant int& batch_ndims [[buffer(15)]],
    -
    1499 const constant int* batch_shape [[buffer(16)]],
    -
    1500 const device uint32_t* lhs_indices [[buffer(17)]],
    -
    1501 const device uint32_t* rhs_indices [[buffer(18)]],
    -
    1502 const constant size_t* lhs_strides [[buffer(19)]],
    -
    1503 const constant size_t* rhs_strides [[buffer(20)]],
    -
    1504 uint3 tid [[threadgroup_position_in_grid]],
    -
    1505 uint simd_gid [[simdgroup_index_in_threadgroup]],
    -
    1506 uint simd_lid [[thread_index_in_simdgroup]]) {
    -
    1507 adjust_matrix_offsets<T>(
    -
    1508 x,
    -
    1509 w,
    -
    1510 scales,
    -
    1511 biases,
    -
    1512 lhs_indices,
    -
    1513 rhs_indices,
    -
    1514 y,
    -
    1515 out_vec_size,
    -
    1516 batch_ndims,
    -
    1517 batch_shape,
    -
    1518 lhs_strides,
    -
    1519 rhs_strides,
    -
    1520 x_batch_ndims,
    -
    1521 x_shape,
    -
    1522 x_strides,
    -
    1523 w_batch_ndims,
    -
    1524 w_shape,
    -
    1525 w_strides,
    -
    1526 s_strides,
    -
    1527 b_strides,
    -
    1528 tid);
    -
    1529 qmv_impl<T, group_size, bits>(
    -
    1530 w,
    -
    1531 scales,
    -
    1532 biases,
    -
    1533 x,
    -
    1534 y,
    -
    1535 in_vec_size,
    -
    1536 out_vec_size,
    -
    1537 tid,
    -
    1538 simd_gid,
    -
    1539 simd_lid);
    -
    1540}
    +
    1474
    +
    1475template <typename T, int group_size, int bits>
    +
    +
    1476[[kernel]] void bs_qmv_fast(
    +
    1477 const device uint32_t* w [[buffer(0)]],
    +
    1478 const device T* scales [[buffer(1)]],
    +
    1479 const device T* biases [[buffer(2)]],
    +
    1480 const device T* x [[buffer(3)]],
    +
    1481 device T* y [[buffer(4)]],
    +
    1482 const constant int& in_vec_size [[buffer(5)]],
    +
    1483 const constant int& out_vec_size [[buffer(6)]],
    +
    1484 const constant int& x_batch_ndims [[buffer(7)]],
    +
    1485 const constant int* x_shape [[buffer(8)]],
    +
    1486 const constant size_t* x_strides [[buffer(9)]],
    +
    1487 const constant int& w_batch_ndims [[buffer(10)]],
    +
    1488 const constant int* w_shape [[buffer(11)]],
    +
    1489 const constant size_t* w_strides [[buffer(12)]],
    +
    1490 const constant size_t* s_strides [[buffer(13)]],
    +
    1491 const constant size_t* b_strides [[buffer(14)]],
    +
    1492 const constant int& batch_ndims [[buffer(15)]],
    +
    1493 const constant int* batch_shape [[buffer(16)]],
    +
    1494 const device uint32_t* lhs_indices [[buffer(17)]],
    +
    1495 const device uint32_t* rhs_indices [[buffer(18)]],
    +
    1496 const constant size_t* lhs_strides [[buffer(19)]],
    +
    1497 const constant size_t* rhs_strides [[buffer(20)]],
    +
    1498 uint3 tid [[threadgroup_position_in_grid]],
    +
    1499 uint simd_gid [[simdgroup_index_in_threadgroup]],
    +
    1500 uint simd_lid [[thread_index_in_simdgroup]]) {
    +
    1501 adjust_matrix_offsets<T>(
    +
    1502 x,
    +
    1503 w,
    +
    1504 scales,
    +
    1505 biases,
    +
    1506 lhs_indices,
    +
    1507 rhs_indices,
    +
    1508 y,
    +
    1509 out_vec_size,
    +
    1510 batch_ndims,
    +
    1511 batch_shape,
    +
    1512 lhs_strides,
    +
    1513 rhs_strides,
    +
    1514 x_batch_ndims,
    +
    1515 x_shape,
    +
    1516 x_strides,
    +
    1517 w_batch_ndims,
    +
    1518 w_shape,
    +
    1519 w_strides,
    +
    1520 s_strides,
    +
    1521 b_strides,
    +
    1522 tid);
    +
    1523 qmv_fast_impl<T, group_size, bits>(
    +
    1524 w,
    +
    1525 scales,
    +
    1526 biases,
    +
    1527 x,
    +
    1528 y,
    +
    1529 in_vec_size,
    +
    1530 out_vec_size,
    +
    1531 tid,
    +
    1532 simd_gid,
    +
    1533 simd_lid);
    +
    1534}
    -
    1541
    -
    1542template <typename T, int group_size, int bits>
    -
    -
    1543[[kernel]] void bs_qvm(
    -
    1544 const device uint32_t* w [[buffer(0)]],
    -
    1545 const device T* scales [[buffer(1)]],
    -
    1546 const device T* biases [[buffer(2)]],
    -
    1547 const device T* x [[buffer(3)]],
    -
    1548 device T* y [[buffer(4)]],
    -
    1549 const constant int& in_vec_size [[buffer(5)]],
    -
    1550 const constant int& out_vec_size [[buffer(6)]],
    -
    1551 const constant int& x_batch_ndims [[buffer(7)]],
    -
    1552 const constant int* x_shape [[buffer(8)]],
    -
    1553 const constant size_t* x_strides [[buffer(9)]],
    -
    1554 const constant int& w_batch_ndims [[buffer(10)]],
    -
    1555 const constant int* w_shape [[buffer(11)]],
    -
    1556 const constant size_t* w_strides [[buffer(12)]],
    -
    1557 const constant size_t* s_strides [[buffer(13)]],
    -
    1558 const constant size_t* b_strides [[buffer(14)]],
    -
    1559 const constant int& batch_ndims [[buffer(15)]],
    -
    1560 const constant int* batch_shape [[buffer(16)]],
    -
    1561 const device uint32_t* lhs_indices [[buffer(17)]],
    -
    1562 const device uint32_t* rhs_indices [[buffer(18)]],
    -
    1563 const constant size_t* lhs_strides [[buffer(19)]],
    -
    1564 const constant size_t* rhs_strides [[buffer(20)]],
    -
    1565 uint3 tid [[threadgroup_position_in_grid]],
    -
    1566 uint simd_gid [[simdgroup_index_in_threadgroup]],
    -
    1567 uint simd_lid [[thread_index_in_simdgroup]]) {
    -
    1568 adjust_matrix_offsets<T>(
    -
    1569 x,
    -
    1570 w,
    -
    1571 scales,
    -
    1572 biases,
    -
    1573 lhs_indices,
    -
    1574 rhs_indices,
    -
    1575 y,
    -
    1576 out_vec_size,
    -
    1577 batch_ndims,
    -
    1578 batch_shape,
    -
    1579 lhs_strides,
    -
    1580 rhs_strides,
    -
    1581 x_batch_ndims,
    -
    1582 x_shape,
    -
    1583 x_strides,
    -
    1584 w_batch_ndims,
    -
    1585 w_shape,
    -
    1586 w_strides,
    -
    1587 s_strides,
    -
    1588 b_strides,
    -
    1589 tid);
    -
    1590 qvm_impl<T, group_size, bits>(
    -
    1591 w,
    -
    1592 scales,
    -
    1593 biases,
    -
    1594 x,
    -
    1595 y,
    -
    1596 in_vec_size,
    -
    1597 out_vec_size,
    -
    1598 tid,
    -
    1599 simd_gid,
    -
    1600 simd_lid);
    -
    1601}
    +
    1535
    +
    1536template <typename T, int group_size, int bits>
    +
    +
    1537[[kernel]] void bs_qmv(
    +
    1538 const device uint32_t* w [[buffer(0)]],
    +
    1539 const device T* scales [[buffer(1)]],
    +
    1540 const device T* biases [[buffer(2)]],
    +
    1541 const device T* x [[buffer(3)]],
    +
    1542 device T* y [[buffer(4)]],
    +
    1543 const constant int& in_vec_size [[buffer(5)]],
    +
    1544 const constant int& out_vec_size [[buffer(6)]],
    +
    1545 const constant int& x_batch_ndims [[buffer(7)]],
    +
    1546 const constant int* x_shape [[buffer(8)]],
    +
    1547 const constant size_t* x_strides [[buffer(9)]],
    +
    1548 const constant int& w_batch_ndims [[buffer(10)]],
    +
    1549 const constant int* w_shape [[buffer(11)]],
    +
    1550 const constant size_t* w_strides [[buffer(12)]],
    +
    1551 const constant size_t* s_strides [[buffer(13)]],
    +
    1552 const constant size_t* b_strides [[buffer(14)]],
    +
    1553 const constant int& batch_ndims [[buffer(15)]],
    +
    1554 const constant int* batch_shape [[buffer(16)]],
    +
    1555 const device uint32_t* lhs_indices [[buffer(17)]],
    +
    1556 const device uint32_t* rhs_indices [[buffer(18)]],
    +
    1557 const constant size_t* lhs_strides [[buffer(19)]],
    +
    1558 const constant size_t* rhs_strides [[buffer(20)]],
    +
    1559 uint3 tid [[threadgroup_position_in_grid]],
    +
    1560 uint simd_gid [[simdgroup_index_in_threadgroup]],
    +
    1561 uint simd_lid [[thread_index_in_simdgroup]]) {
    +
    1562 adjust_matrix_offsets<T>(
    +
    1563 x,
    +
    1564 w,
    +
    1565 scales,
    +
    1566 biases,
    +
    1567 lhs_indices,
    +
    1568 rhs_indices,
    +
    1569 y,
    +
    1570 out_vec_size,
    +
    1571 batch_ndims,
    +
    1572 batch_shape,
    +
    1573 lhs_strides,
    +
    1574 rhs_strides,
    +
    1575 x_batch_ndims,
    +
    1576 x_shape,
    +
    1577 x_strides,
    +
    1578 w_batch_ndims,
    +
    1579 w_shape,
    +
    1580 w_strides,
    +
    1581 s_strides,
    +
    1582 b_strides,
    +
    1583 tid);
    +
    1584 qmv_impl<T, group_size, bits>(
    +
    1585 w,
    +
    1586 scales,
    +
    1587 biases,
    +
    1588 x,
    +
    1589 y,
    +
    1590 in_vec_size,
    +
    1591 out_vec_size,
    +
    1592 tid,
    +
    1593 simd_gid,
    +
    1594 simd_lid);
    +
    1595}
    -
    1602
    -
    1603template <
    -
    1604 typename T,
    -
    1605 const int group_size,
    -
    1606 const int bits,
    -
    1607 const bool aligned_N,
    -
    1608 const int BM = 32,
    -
    1609 const int BK = 32,
    -
    1610 const int BN = 32>
    -
    -
    1611[[kernel]] void bs_qmm_t(
    -
    1612 const device uint32_t* w [[buffer(0)]],
    -
    1613 const device T* scales [[buffer(1)]],
    -
    1614 const device T* biases [[buffer(2)]],
    -
    1615 const device T* x [[buffer(3)]],
    -
    1616 device T* y [[buffer(4)]],
    -
    1617 const constant int& K [[buffer(5)]],
    -
    1618 const constant int& N [[buffer(6)]],
    -
    1619 const constant int& M [[buffer(7)]],
    -
    1620 const constant int& x_batch_ndims [[buffer(8)]],
    -
    1621 const constant int* x_shape [[buffer(9)]],
    -
    1622 const constant size_t* x_strides [[buffer(10)]],
    -
    1623 const constant int& w_batch_ndims [[buffer(11)]],
    -
    1624 const constant int* w_shape [[buffer(12)]],
    -
    1625 const constant size_t* w_strides [[buffer(13)]],
    -
    1626 const constant size_t* s_strides [[buffer(14)]],
    -
    1627 const constant size_t* b_strides [[buffer(15)]],
    -
    1628 const constant int& batch_ndims [[buffer(16)]],
    -
    1629 const constant int* batch_shape [[buffer(17)]],
    -
    1630 const device uint32_t* lhs_indices [[buffer(18)]],
    -
    1631 const device uint32_t* rhs_indices [[buffer(19)]],
    -
    1632 const constant size_t* lhs_strides [[buffer(20)]],
    -
    1633 const constant size_t* rhs_strides [[buffer(21)]],
    -
    1634 uint3 tid [[threadgroup_position_in_grid]],
    -
    1635 uint lid [[thread_index_in_threadgroup]],
    -
    1636 uint simd_gid [[simdgroup_index_in_threadgroup]],
    -
    1637 uint simd_lid [[thread_index_in_simdgroup]]) {
    -
    1638 (void)lid;
    -
    1639
    -
    1640 constexpr int BK_padded = (BK + 16 / sizeof(T));
    -
    1641
    -
    1642 threadgroup T Xs[BM * BK_padded];
    -
    1643 threadgroup T Ws[BN * BK_padded];
    -
    1644
    -
    1645 adjust_matrix_offsets<T>(
    -
    1646 x,
    -
    1647 w,
    -
    1648 scales,
    -
    1649 biases,
    -
    1650 lhs_indices,
    -
    1651 rhs_indices,
    -
    1652 y,
    -
    1653 M * N,
    -
    1654 batch_ndims,
    -
    1655 batch_shape,
    -
    1656 lhs_strides,
    -
    1657 rhs_strides,
    -
    1658 x_batch_ndims,
    -
    1659 x_shape,
    -
    1660 x_strides,
    -
    1661 w_batch_ndims,
    -
    1662 w_shape,
    -
    1663 w_strides,
    -
    1664 s_strides,
    -
    1665 b_strides,
    -
    1666 tid);
    -
    1667 qmm_t_impl<T, group_size, bits, aligned_N, BM, BK, BN>(
    -
    1668 w, scales, biases, x, y, Xs, Ws, K, N, M, tid, lid, simd_gid, simd_lid);
    -
    1669}
    +
    1596
    +
    1597template <typename T, int group_size, int bits>
    +
    +
    1598[[kernel]] void bs_qvm(
    +
    1599 const device uint32_t* w [[buffer(0)]],
    +
    1600 const device T* scales [[buffer(1)]],
    +
    1601 const device T* biases [[buffer(2)]],
    +
    1602 const device T* x [[buffer(3)]],
    +
    1603 device T* y [[buffer(4)]],
    +
    1604 const constant int& in_vec_size [[buffer(5)]],
    +
    1605 const constant int& out_vec_size [[buffer(6)]],
    +
    1606 const constant int& x_batch_ndims [[buffer(7)]],
    +
    1607 const constant int* x_shape [[buffer(8)]],
    +
    1608 const constant size_t* x_strides [[buffer(9)]],
    +
    1609 const constant int& w_batch_ndims [[buffer(10)]],
    +
    1610 const constant int* w_shape [[buffer(11)]],
    +
    1611 const constant size_t* w_strides [[buffer(12)]],
    +
    1612 const constant size_t* s_strides [[buffer(13)]],
    +
    1613 const constant size_t* b_strides [[buffer(14)]],
    +
    1614 const constant int& batch_ndims [[buffer(15)]],
    +
    1615 const constant int* batch_shape [[buffer(16)]],
    +
    1616 const device uint32_t* lhs_indices [[buffer(17)]],
    +
    1617 const device uint32_t* rhs_indices [[buffer(18)]],
    +
    1618 const constant size_t* lhs_strides [[buffer(19)]],
    +
    1619 const constant size_t* rhs_strides [[buffer(20)]],
    +
    1620 uint3 tid [[threadgroup_position_in_grid]],
    +
    1621 uint simd_gid [[simdgroup_index_in_threadgroup]],
    +
    1622 uint simd_lid [[thread_index_in_simdgroup]]) {
    +
    1623 adjust_matrix_offsets<T>(
    +
    1624 x,
    +
    1625 w,
    +
    1626 scales,
    +
    1627 biases,
    +
    1628 lhs_indices,
    +
    1629 rhs_indices,
    +
    1630 y,
    +
    1631 out_vec_size,
    +
    1632 batch_ndims,
    +
    1633 batch_shape,
    +
    1634 lhs_strides,
    +
    1635 rhs_strides,
    +
    1636 x_batch_ndims,
    +
    1637 x_shape,
    +
    1638 x_strides,
    +
    1639 w_batch_ndims,
    +
    1640 w_shape,
    +
    1641 w_strides,
    +
    1642 s_strides,
    +
    1643 b_strides,
    +
    1644 tid);
    +
    1645 qvm_impl<T, group_size, bits>(
    +
    1646 w,
    +
    1647 scales,
    +
    1648 biases,
    +
    1649 x,
    +
    1650 y,
    +
    1651 in_vec_size,
    +
    1652 out_vec_size,
    +
    1653 tid,
    +
    1654 simd_gid,
    +
    1655 simd_lid);
    +
    1656}
    -
    1670
    -
    1671template <
    -
    1672 typename T,
    -
    1673 const int group_size,
    -
    1674 const int bits,
    -
    1675 const int BM = 32,
    -
    1676 const int BK = 32,
    -
    1677 const int BN = 32>
    -
    -
    1678[[kernel]] void bs_qmm_n(
    -
    1679 const device uint32_t* w [[buffer(0)]],
    -
    1680 const device T* scales [[buffer(1)]],
    -
    1681 const device T* biases [[buffer(2)]],
    -
    1682 const device T* x [[buffer(3)]],
    -
    1683 device T* y [[buffer(4)]],
    -
    1684 const constant int& K [[buffer(5)]],
    -
    1685 const constant int& N [[buffer(6)]],
    -
    1686 const constant int& M [[buffer(7)]],
    -
    1687 const constant int& x_batch_ndims [[buffer(8)]],
    -
    1688 const constant int* x_shape [[buffer(9)]],
    -
    1689 const constant size_t* x_strides [[buffer(10)]],
    -
    1690 const constant int& w_batch_ndims [[buffer(11)]],
    -
    1691 const constant int* w_shape [[buffer(12)]],
    -
    1692 const constant size_t* w_strides [[buffer(13)]],
    -
    1693 const constant size_t* s_strides [[buffer(14)]],
    -
    1694 const constant size_t* b_strides [[buffer(15)]],
    -
    1695 const constant int& batch_ndims [[buffer(16)]],
    -
    1696 const constant int* batch_shape [[buffer(17)]],
    -
    1697 const device uint32_t* lhs_indices [[buffer(18)]],
    -
    1698 const device uint32_t* rhs_indices [[buffer(19)]],
    -
    1699 const constant size_t* lhs_strides [[buffer(20)]],
    -
    1700 const constant size_t* rhs_strides [[buffer(21)]],
    -
    1701 uint3 tid [[threadgroup_position_in_grid]],
    -
    1702 uint lid [[thread_index_in_threadgroup]],
    -
    1703 uint simd_gid [[simdgroup_index_in_threadgroup]],
    -
    1704 uint simd_lid [[thread_index_in_simdgroup]]) {
    -
    1705 (void)lid;
    -
    1706
    -
    1707 constexpr int BK_padded = (BK + 16 / sizeof(T));
    -
    1708 constexpr int BN_padded = (BN + 16 / sizeof(T));
    -
    1709
    -
    1710 threadgroup T Xs[BM * BK_padded];
    -
    1711 threadgroup T Ws[BK * BN_padded];
    -
    1712
    -
    1713 adjust_matrix_offsets<T>(
    -
    1714 x,
    -
    1715 w,
    -
    1716 scales,
    -
    1717 biases,
    -
    1718 lhs_indices,
    -
    1719 rhs_indices,
    -
    1720 y,
    -
    1721 M * N,
    -
    1722 batch_ndims,
    -
    1723 batch_shape,
    -
    1724 lhs_strides,
    -
    1725 rhs_strides,
    -
    1726 x_batch_ndims,
    -
    1727 x_shape,
    -
    1728 x_strides,
    -
    1729 w_batch_ndims,
    -
    1730 w_shape,
    -
    1731 w_strides,
    -
    1732 s_strides,
    -
    1733 b_strides,
    -
    1734 tid);
    -
    1735 qmm_n_impl<T, group_size, bits, BM, BK, BN>(
    -
    1736 w, scales, biases, x, y, Xs, Ws, K, N, M, tid, lid, simd_gid, simd_lid);
    -
    1737}
    +
    1657
    +
    1658template <
    +
    1659 typename T,
    +
    1660 const int group_size,
    +
    1661 const int bits,
    +
    1662 const bool aligned_N,
    +
    1663 const int BM = 32,
    +
    1664 const int BK = 32,
    +
    1665 const int BN = 32>
    +
    +
    1666[[kernel]] void bs_qmm_t(
    +
    1667 const device uint32_t* w [[buffer(0)]],
    +
    1668 const device T* scales [[buffer(1)]],
    +
    1669 const device T* biases [[buffer(2)]],
    +
    1670 const device T* x [[buffer(3)]],
    +
    1671 device T* y [[buffer(4)]],
    +
    1672 const constant int& K [[buffer(5)]],
    +
    1673 const constant int& N [[buffer(6)]],
    +
    1674 const constant int& M [[buffer(7)]],
    +
    1675 const constant int& x_batch_ndims [[buffer(8)]],
    +
    1676 const constant int* x_shape [[buffer(9)]],
    +
    1677 const constant size_t* x_strides [[buffer(10)]],
    +
    1678 const constant int& w_batch_ndims [[buffer(11)]],
    +
    1679 const constant int* w_shape [[buffer(12)]],
    +
    1680 const constant size_t* w_strides [[buffer(13)]],
    +
    1681 const constant size_t* s_strides [[buffer(14)]],
    +
    1682 const constant size_t* b_strides [[buffer(15)]],
    +
    1683 const constant int& batch_ndims [[buffer(16)]],
    +
    1684 const constant int* batch_shape [[buffer(17)]],
    +
    1685 const device uint32_t* lhs_indices [[buffer(18)]],
    +
    1686 const device uint32_t* rhs_indices [[buffer(19)]],
    +
    1687 const constant size_t* lhs_strides [[buffer(20)]],
    +
    1688 const constant size_t* rhs_strides [[buffer(21)]],
    +
    1689 uint3 tid [[threadgroup_position_in_grid]],
    +
    1690 uint lid [[thread_index_in_threadgroup]],
    +
    1691 uint simd_gid [[simdgroup_index_in_threadgroup]],
    +
    1692 uint simd_lid [[thread_index_in_simdgroup]]) {
    +
    1693 (void)lid;
    +
    1694
    +
    1695 constexpr int BK_padded = (BK + 16 / sizeof(T));
    +
    1696
    +
    1697 threadgroup T Xs[BM * BK_padded];
    +
    1698 threadgroup T Ws[BN * BK_padded];
    +
    1699
    +
    1700 adjust_matrix_offsets<T>(
    +
    1701 x,
    +
    1702 w,
    +
    1703 scales,
    +
    1704 biases,
    +
    1705 lhs_indices,
    +
    1706 rhs_indices,
    +
    1707 y,
    +
    1708 M * N,
    +
    1709 batch_ndims,
    +
    1710 batch_shape,
    +
    1711 lhs_strides,
    +
    1712 rhs_strides,
    +
    1713 x_batch_ndims,
    +
    1714 x_shape,
    +
    1715 x_strides,
    +
    1716 w_batch_ndims,
    +
    1717 w_shape,
    +
    1718 w_strides,
    +
    1719 s_strides,
    +
    1720 b_strides,
    +
    1721 tid);
    +
    1722 qmm_t_impl<T, group_size, bits, aligned_N, BM, BK, BN>(
    +
    1723 w, scales, biases, x, y, Xs, Ws, K, N, M, tid, lid, simd_gid, simd_lid);
    +
    1724}
    -
    1738
    -
    1739template <typename T, const int group_size, const int bits>
    -
    -
    1740[[kernel]] void affine_quantize(
    -
    1741 const device T* w [[buffer(0)]],
    -
    1742 device uint8_t* out [[buffer(1)]],
    -
    1743 device T* scales [[buffer(2)]],
    -
    1744 device T* biases [[buffer(3)]],
    -
    1745 uint2 index [[thread_position_in_grid]],
    -
    1746 uint2 grid_dim [[threads_per_grid]]) {
    -
    1747 constexpr T eps = T(1e-7);
    -
    1748 constexpr int simd_size = 32;
    -
    1749 constexpr int uint8_bits = 8;
    -
    1750 constexpr T n_bins = (1 << bits) - 1;
    -
    1751 constexpr int packs_per_int = uint8_bits / bits;
    -
    1752 constexpr int values_per_reduce = group_size / simd_size;
    -
    1753 constexpr int writes_per_reduce = packs_per_int / values_per_reduce;
    -
    1754 constexpr int writes_per_pack =
    -
    1755 writes_per_reduce > 1 ? 1 : values_per_reduce / packs_per_int;
    -
    1756
    -
    1757 static_assert(
    -
    1758 group_size % simd_size == 0,
    -
    1759 "Group size must be divisible by simd size.");
    -
    1760
    -
    1761 size_t offset = index.x + grid_dim.x * size_t(index.y);
    -
    1762 size_t in_index = offset * values_per_reduce;
    -
    1763 size_t out_index = offset * writes_per_pack;
    +
    1725
    +
    1726template <
    +
    1727 typename T,
    +
    1728 const int group_size,
    +
    1729 const int bits,
    +
    1730 const int BM = 32,
    +
    1731 const int BK = 32,
    +
    1732 const int BN = 32>
    +
    +
    1733[[kernel]] void bs_qmm_n(
    +
    1734 const device uint32_t* w [[buffer(0)]],
    +
    1735 const device T* scales [[buffer(1)]],
    +
    1736 const device T* biases [[buffer(2)]],
    +
    1737 const device T* x [[buffer(3)]],
    +
    1738 device T* y [[buffer(4)]],
    +
    1739 const constant int& K [[buffer(5)]],
    +
    1740 const constant int& N [[buffer(6)]],
    +
    1741 const constant int& M [[buffer(7)]],
    +
    1742 const constant int& x_batch_ndims [[buffer(8)]],
    +
    1743 const constant int* x_shape [[buffer(9)]],
    +
    1744 const constant size_t* x_strides [[buffer(10)]],
    +
    1745 const constant int& w_batch_ndims [[buffer(11)]],
    +
    1746 const constant int* w_shape [[buffer(12)]],
    +
    1747 const constant size_t* w_strides [[buffer(13)]],
    +
    1748 const constant size_t* s_strides [[buffer(14)]],
    +
    1749 const constant size_t* b_strides [[buffer(15)]],
    +
    1750 const constant int& batch_ndims [[buffer(16)]],
    +
    1751 const constant int* batch_shape [[buffer(17)]],
    +
    1752 const device uint32_t* lhs_indices [[buffer(18)]],
    +
    1753 const device uint32_t* rhs_indices [[buffer(19)]],
    +
    1754 const constant size_t* lhs_strides [[buffer(20)]],
    +
    1755 const constant size_t* rhs_strides [[buffer(21)]],
    +
    1756 uint3 tid [[threadgroup_position_in_grid]],
    +
    1757 uint lid [[thread_index_in_threadgroup]],
    +
    1758 uint simd_gid [[simdgroup_index_in_threadgroup]],
    +
    1759 uint simd_lid [[thread_index_in_simdgroup]]) {
    +
    1760 (void)lid;
    +
    1761
    +
    1762 constexpr int BK_padded = (BK + 16 / sizeof(T));
    +
    1763 constexpr int BN_padded = (BN + 16 / sizeof(T));
    1764
    -
    1765 T w_thread[values_per_reduce];
    -
    1766 T w_min = Limits<T>::max;
    -
    1767 T w_max = 0;
    -
    1768
    -
    1769#pragma clang loop unroll(full)
    -
    1770 for (int i = 0; i < values_per_reduce; i++) {
    -
    1771 T val = w[in_index + i];
    -
    1772 w_thread[i] = val;
    -
    1773 w_min = min(w_min, val);
    -
    1774 w_max = max(w_max, val);
    -
    1775 }
    -
    1776
    -
    1777 w_min = simd_min(w_min);
    -
    1778 w_max = simd_max(w_max);
    -
    1779
    -
    1780 T scale = max((w_max - w_min) / n_bins, eps);
    -
    1781 bool side = abs(w_min) > abs(w_max);
    -
    1782 scale = side ? scale : -scale;
    -
    1783 T edge = side ? w_min : w_max;
    -
    1784 T q0 = round(edge / scale);
    -
    1785 bool at_zero = q0 == 0.0f;
    -
    1786 scale = at_zero ? scale : edge / q0;
    -
    1787 T bias = at_zero ? T(0) : edge;
    -
    1788
    -
    1789 // Write out the scales and biases
    -
    1790 size_t gindex = in_index / group_size;
    -
    1791 if (in_index % group_size == 0) {
    -
    1792 scales[gindex] = scale;
    -
    1793 biases[gindex] = bias;
    -
    1794 }
    -
    1795
    -
    1796 uint8_t output = 0;
    -
    1797#pragma clang loop unroll(full)
    -
    1798 for (int i = 0; i < values_per_reduce; i++) {
    -
    1799 uint8_t val = min(round((w_thread[i] - bias) / scale), n_bins);
    -
    1800 if (bits == 8) {
    -
    1801 output = val;
    -
    1802 } else {
    -
    1803 output += val << (bits * (i % packs_per_int));
    -
    1804 }
    -
    1805
    -
    1806 if (packs_per_int < values_per_reduce &&
    -
    1807 i % packs_per_int == packs_per_int - 1) {
    -
    1808 out[out_index + i / packs_per_int] = output;
    -
    1809 output = 0;
    -
    1810 } else {
    -
    1811#pragma clang loop unroll(full)
    -
    1812 for (int j = 0; j < writes_per_reduce - 1; j++) {
    -
    1813 uint8_t sval = simd_shuffle_down(val, j + 1);
    -
    1814 output += sval << (bits * (values_per_reduce + j + i));
    -
    1815 }
    -
    1816 }
    -
    1817 }
    -
    1818 if (writes_per_reduce > 0 && out_index % writes_per_reduce == 0) {
    -
    1819 out[out_index / writes_per_reduce] = output;
    -
    1820 }
    -
    1821}
    +
    1765 threadgroup T Xs[BM * BK_padded];
    +
    1766 threadgroup T Ws[BK * BN_padded];
    +
    1767
    +
    1768 adjust_matrix_offsets<T>(
    +
    1769 x,
    +
    1770 w,
    +
    1771 scales,
    +
    1772 biases,
    +
    1773 lhs_indices,
    +
    1774 rhs_indices,
    +
    1775 y,
    +
    1776 M * N,
    +
    1777 batch_ndims,
    +
    1778 batch_shape,
    +
    1779 lhs_strides,
    +
    1780 rhs_strides,
    +
    1781 x_batch_ndims,
    +
    1782 x_shape,
    +
    1783 x_strides,
    +
    1784 w_batch_ndims,
    +
    1785 w_shape,
    +
    1786 w_strides,
    +
    1787 s_strides,
    +
    1788 b_strides,
    +
    1789 tid);
    +
    1790 qmm_n_impl<T, group_size, bits, BM, BK, BN>(
    +
    1791 w, scales, biases, x, y, Xs, Ws, K, N, M, tid, lid, simd_gid, simd_lid);
    +
    1792}
    -
    1822
    -
    1823template <typename T, const int group_size, const int bits>
    -
    -
    1824[[kernel]] void affine_quantize_scales_biases(
    -
    1825 const device T* w [[buffer(0)]],
    -
    1826 const device T* scales [[buffer(1)]],
    -
    1827 const device T* biases [[buffer(2)]],
    -
    1828 device uint8_t* out [[buffer(3)]],
    -
    1829 uint2 index [[thread_position_in_grid]],
    -
    1830 uint2 grid_dim [[threads_per_grid]]) {
    -
    1831 constexpr int uint8_bits = 8;
    -
    1832 constexpr int packs_per_int = uint8_bits / bits;
    -
    1833 constexpr T n_bins = (1 << bits) - 1;
    +
    1793
    +
    1794template <typename T, const int group_size, const int bits>
    +
    +
    1795[[kernel]] void affine_quantize(
    +
    1796 const device T* w [[buffer(0)]],
    +
    1797 device uint8_t* out [[buffer(1)]],
    +
    1798 device T* scales [[buffer(2)]],
    +
    1799 device T* biases [[buffer(3)]],
    +
    1800 uint2 index [[thread_position_in_grid]],
    +
    1801 uint2 grid_dim [[threads_per_grid]]) {
    +
    1802 constexpr T eps = T(1e-7);
    +
    1803 constexpr int simd_size = 32;
    +
    1804 constexpr int uint8_bits = 8;
    +
    1805 constexpr T n_bins = (1 << bits) - 1;
    +
    1806 constexpr int packs_per_int = uint8_bits / bits;
    +
    1807 constexpr int values_per_reduce = group_size / simd_size;
    +
    1808 constexpr int writes_per_reduce = packs_per_int / values_per_reduce;
    +
    1809 constexpr int writes_per_pack =
    +
    1810 writes_per_reduce > 1 ? 1 : values_per_reduce / packs_per_int;
    +
    1811
    +
    1812 static_assert(
    +
    1813 group_size % simd_size == 0,
    +
    1814 "Group size must be divisible by simd size.");
    +
    1815
    +
    1816 size_t offset = index.x + grid_dim.x * size_t(index.y);
    +
    1817 size_t in_index = offset * values_per_reduce;
    +
    1818 size_t out_index = offset * writes_per_pack;
    +
    1819
    +
    1820 T w_thread[values_per_reduce];
    +
    1821 T w_min = Limits<T>::max;
    +
    1822 T w_max = 0;
    +
    1823
    +
    1824#pragma clang loop unroll(full)
    +
    1825 for (int i = 0; i < values_per_reduce; i++) {
    +
    1826 T val = w[in_index + i];
    +
    1827 w_thread[i] = val;
    +
    1828 w_min = min(w_min, val);
    +
    1829 w_max = max(w_max, val);
    +
    1830 }
    +
    1831
    +
    1832 w_min = simd_min(w_min);
    +
    1833 w_max = simd_max(w_max);
    1834
    -
    1835 size_t offset = index.x + grid_dim.x * size_t(index.y);
    -
    1836 size_t in_index = offset * packs_per_int;
    -
    1837 size_t gindex = in_index / group_size;
    -
    1838
    -
    1839 T scale = scales[gindex];
    -
    1840 T bias = biases[gindex];
    -
    1841
    -
    1842 uint8_t output = 0;
    -
    1843#pragma clang loop unroll(full)
    -
    1844 for (int i = 0; i < packs_per_int; i++) {
    -
    1845 uint8_t val = min(round((w[in_index + i] - bias) / scale), n_bins);
    -
    1846 if (bits == 8) {
    -
    1847 output = val;
    -
    1848 } else {
    -
    1849 output += val << (bits * i);
    -
    1850 }
    -
    1851 }
    -
    1852 out[offset] = output;
    -
    1853}
    +
    1835 T scale = max((w_max - w_min) / n_bins, eps);
    +
    1836 bool side = abs(w_min) > abs(w_max);
    +
    1837 scale = side ? scale : -scale;
    +
    1838 T edge = side ? w_min : w_max;
    +
    1839 T q0 = round(edge / scale);
    +
    1840 bool at_zero = q0 == 0.0f;
    +
    1841 scale = at_zero ? scale : edge / q0;
    +
    1842 T bias = at_zero ? T(0) : edge;
    +
    1843
    +
    1844 // Write out the scales and biases
    +
    1845 size_t gindex = in_index / group_size;
    +
    1846 if (in_index % group_size == 0) {
    +
    1847 scales[gindex] = scale;
    +
    1848 biases[gindex] = bias;
    +
    1849 }
    +
    1850
    +
    1851 uint8_t output = 0;
    +
    1852#pragma clang loop unroll(full)
    +
    1853 for (int i = 0; i < values_per_reduce; i++) {
    +
    1854 uint8_t val = min(round((w_thread[i] - bias) / scale), n_bins);
    +
    1855 if (bits == 8) {
    +
    1856 output = val;
    +
    1857 } else {
    +
    1858 output += val << (bits * (i % packs_per_int));
    +
    1859 }
    +
    1860
    +
    1861 if (packs_per_int < values_per_reduce &&
    +
    1862 i % packs_per_int == packs_per_int - 1) {
    +
    1863 out[out_index + i / packs_per_int] = output;
    +
    1864 output = 0;
    +
    1865 } else {
    +
    1866#pragma clang loop unroll(full)
    +
    1867 for (int j = 0; j < writes_per_reduce - 1; j++) {
    +
    1868 uint8_t sval = simd_shuffle_down(val, j + 1);
    +
    1869 output += sval << (bits * (values_per_reduce + j + i));
    +
    1870 }
    +
    1871 }
    +
    1872 }
    +
    1873 if (writes_per_reduce > 0 && out_index % writes_per_reduce == 0) {
    +
    1874 out[out_index / writes_per_reduce] = output;
    +
    1875 }
    +
    1876}
    -
    1854
    -
    1855template <typename T, const int group_size, const int bits>
    -
    -
    1856[[kernel]] void affine_dequantize(
    -
    1857 const device uint8_t* w [[buffer(0)]],
    -
    1858 const device T* scales [[buffer(1)]],
    -
    1859 const device T* biases [[buffer(2)]],
    -
    1860 device T* out [[buffer(3)]],
    -
    1861 uint2 index [[thread_position_in_grid]],
    -
    1862 uint2 grid_dim [[threads_per_grid]]) {
    -
    1863 constexpr int uint8_bits = 8;
    -
    1864 constexpr int packs_per_int = uint8_bits / bits;
    -
    1865
    -
    1866 size_t offset = index.x + grid_dim.x * size_t(index.y);
    -
    1867 size_t oindex = offset * packs_per_int;
    -
    1868 size_t gindex = oindex / group_size;
    -
    1869 T scale = scales[gindex];
    -
    1870 T bias = biases[gindex];
    -
    1871 uint val = w[offset];
    -
    1872
    -
    1873#pragma clang loop unroll(full)
    -
    1874 for (int i = 0; i < packs_per_int; i++) {
    -
    1875 uint8_t d;
    -
    1876 if (bits == 2) {
    -
    1877 d = (val >> (bits * i)) & 0x03;
    -
    1878 } else if (bits == 4) {
    -
    1879 d = (val >> (bits * i)) & 0x0f;
    -
    1880 } else if (bits == 8) {
    -
    1881 d = val;
    -
    1882 }
    -
    1883 out[oindex + i] = scale * d + bias;
    -
    1884 }
    -
    1885}
    +
    1877
    +
    1878template <typename T, const int group_size, const int bits>
    +
    +
    1879[[kernel]] void affine_quantize_scales_biases(
    +
    1880 const device T* w [[buffer(0)]],
    +
    1881 const device T* scales [[buffer(1)]],
    +
    1882 const device T* biases [[buffer(2)]],
    +
    1883 device uint8_t* out [[buffer(3)]],
    +
    1884 uint2 index [[thread_position_in_grid]],
    +
    1885 uint2 grid_dim [[threads_per_grid]]) {
    +
    1886 constexpr int uint8_bits = 8;
    +
    1887 constexpr int packs_per_int = uint8_bits / bits;
    +
    1888 constexpr T n_bins = (1 << bits) - 1;
    +
    1889
    +
    1890 size_t offset = index.x + grid_dim.x * size_t(index.y);
    +
    1891 size_t in_index = offset * packs_per_int;
    +
    1892 size_t gindex = in_index / group_size;
    +
    1893
    +
    1894 T scale = scales[gindex];
    +
    1895 T bias = biases[gindex];
    +
    1896
    +
    1897 uint8_t output = 0;
    +
    1898#pragma clang loop unroll(full)
    +
    1899 for (int i = 0; i < packs_per_int; i++) {
    +
    1900 uint8_t val = min(round((w[in_index + i] - bias) / scale), n_bins);
    +
    1901 if (bits == 8) {
    +
    1902 output = val;
    +
    1903 } else {
    +
    1904 output += val << (bits * i);
    +
    1905 }
    +
    1906 }
    +
    1907 out[offset] = output;
    +
    1908}
    +
    +
    1909
    +
    1910template <typename T, const int group_size, const int bits>
    +
    +
    1911[[kernel]] void affine_dequantize(
    +
    1912 const device uint8_t* w [[buffer(0)]],
    +
    1913 const device T* scales [[buffer(1)]],
    +
    1914 const device T* biases [[buffer(2)]],
    +
    1915 device T* out [[buffer(3)]],
    +
    1916 uint2 index [[thread_position_in_grid]],
    +
    1917 uint2 grid_dim [[threads_per_grid]]) {
    +
    1918 constexpr int uint8_bits = 8;
    +
    1919 constexpr int packs_per_int = uint8_bits / bits;
    +
    1920
    +
    1921 size_t offset = index.x + grid_dim.x * size_t(index.y);
    +
    1922 size_t oindex = offset * packs_per_int;
    +
    1923 size_t gindex = oindex / group_size;
    +
    1924 T scale = scales[gindex];
    +
    1925 T bias = biases[gindex];
    +
    1926 uint val = w[offset];
    +
    1927
    +
    1928#pragma clang loop unroll(full)
    +
    1929 for (int i = 0; i < packs_per_int; i++) {
    +
    1930 uint8_t d;
    +
    1931 if (bits == 2) {
    +
    1932 d = (val >> (bits * i)) & 0x03;
    +
    1933 } else if (bits == 4) {
    +
    1934 d = (val >> (bits * i)) & 0x0f;
    +
    1935 } else if (bits == 8) {
    +
    1936 d = val;
    +
    1937 }
    +
    1938 out[oindex + i] = scale * d + bias;
    +
    1939 }
    +
    1940}
    simd_size
    static constant constexpr const uint8_t simd_size
    Definition ops.h:22
    elem_to_loc_broadcast
    METAL_FUNC ulong2 elem_to_loc_broadcast(uint elem, constant const int *shape, constant const size_t *a_strides, constant const size_t *b_strides, int ndim)
    Definition utils.h:7
    @@ -2058,28 +2115,29 @@ $(function(){ initResizable(false); });
    MLX_MTL_CONST
    #define MLX_MTL_CONST
    Definition quantized.h:8
    qdot_safe
    U qdot_safe(const device uint8_t *w, const thread U *x_thread, U scale, U bias, U sum, int N)
    Definition quantized.h:142
    qmm_n_impl
    METAL_FUNC void qmm_n_impl(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, threadgroup T *Xs, threadgroup T *Ws, const constant int &K, const constant int &N, const constant int &M, uint3 tid, uint lid, uint simd_gid, uint simd_lid)
    Definition quantized.h:879
    -
    bs_qmm_n
    void bs_qmm_n(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &K, const constant int &N, const constant int &M, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &batch_ndims, const constant int *batch_shape, const device uint32_t *lhs_indices, const device uint32_t *rhs_indices, const constant size_t *lhs_strides, const constant size_t *rhs_strides, uint3 tid, uint lid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1678
    -
    qmm_n
    void qmm_n(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &K, const constant int &N, const constant int &M, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, uint3 tid, uint lid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1368
    -
    affine_quantize
    void affine_quantize(const device T *w, device uint8_t *out, device T *scales, device T *biases, uint2 index, uint2 grid_dim)
    Definition quantized.h:1740
    -
    qvm_impl
    METAL_FUNC void qvm_impl(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:647
    -
    bs_qmv_fast
    void bs_qmv_fast(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &batch_ndims, const constant int *batch_shape, const device uint32_t *lhs_indices, const device uint32_t *rhs_indices, const constant size_t *lhs_strides, const constant size_t *rhs_strides, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1421
    -
    affine_dequantize
    void affine_dequantize(const device uint8_t *w, const device T *scales, const device T *biases, device T *out, uint2 index, uint2 grid_dim)
    Definition quantized.h:1856
    +
    qvm_impl
    METAL_FUNC void qvm_impl(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const int in_vec_size, const int out_vec_size, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:647
    +
    bs_qmm_n
    void bs_qmm_n(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &K, const constant int &N, const constant int &M, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &batch_ndims, const constant int *batch_shape, const device uint32_t *lhs_indices, const device uint32_t *rhs_indices, const constant size_t *lhs_strides, const constant size_t *rhs_strides, uint3 tid, uint lid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1733
    +
    qmm_n
    void qmm_n(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &K, const constant int &N, const constant int &M, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, uint3 tid, uint lid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1423
    +
    affine_quantize
    void affine_quantize(const device T *w, device uint8_t *out, device T *scales, device T *biases, uint2 index, uint2 grid_dim)
    Definition quantized.h:1795
    +
    bs_qmv_fast
    void bs_qmv_fast(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &batch_ndims, const constant int *batch_shape, const device uint32_t *lhs_indices, const device uint32_t *rhs_indices, const constant size_t *lhs_strides, const constant size_t *rhs_strides, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1476
    +
    affine_dequantize
    void affine_dequantize(const device uint8_t *w, const device T *scales, const device T *biases, device T *out, uint2 index, uint2 grid_dim)
    Definition quantized.h:1911
    SIMD_SIZE
    static constant constexpr const int SIMD_SIZE
    Definition quantized.h:10
    qmv
    void qmv(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1200
    -
    bs_qvm
    void bs_qvm(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &batch_ndims, const constant int *batch_shape, const device uint32_t *lhs_indices, const device uint32_t *rhs_indices, const constant size_t *lhs_strides, const constant size_t *rhs_strides, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1543
    -
    affine_quantize_scales_biases
    void affine_quantize_scales_biases(const device T *w, const device T *scales, const device T *biases, device uint8_t *out, uint2 index, uint2 grid_dim)
    Definition quantized.h:1824
    +
    bs_qvm
    void bs_qvm(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &batch_ndims, const constant int *batch_shape, const device uint32_t *lhs_indices, const device uint32_t *rhs_indices, const constant size_t *lhs_strides, const constant size_t *rhs_strides, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1598
    +
    affine_quantize_scales_biases
    void affine_quantize_scales_biases(const device T *w, const device T *scales, const device T *biases, device uint8_t *out, uint2 index, uint2 grid_dim)
    Definition quantized.h:1879
    qmv_fast
    void qmv_fast(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1149
    qmv_quad
    void qmv_quad(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, uint3 tid, uint quad_gid, uint quad_lid)
    Definition quantized.h:1098
    QUAD_SIZE
    static constant constexpr const int QUAD_SIZE
    Definition quantized.h:11
    load_vector
    U load_vector(const device T *x, thread U *x_thread)
    Definition quantized.h:14
    qmv_impl
    METAL_FUNC void qmv_impl(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:498
    load_vector_safe
    U load_vector_safe(const device T *x, thread U *x_thread, int N)
    Definition quantized.h:52
    -
    bs_qmm_t
    void bs_qmm_t(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &K, const constant int &N, const constant int &M, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &batch_ndims, const constant int *batch_shape, const device uint32_t *lhs_indices, const device uint32_t *rhs_indices, const constant size_t *lhs_strides, const constant size_t *rhs_strides, uint3 tid, uint lid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1611
    +
    bs_qmm_t
    void bs_qmm_t(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &K, const constant int &N, const constant int &M, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &batch_ndims, const constant int *batch_shape, const device uint32_t *lhs_indices, const device uint32_t *rhs_indices, const constant size_t *lhs_strides, const constant size_t *rhs_strides, uint3 tid, uint lid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1666
    qdot
    U qdot(const device uint8_t *w, const thread U *x_thread, U scale, U bias, U sum)
    Definition quantized.h:99
    +
    qvm_split_k
    void qvm_split_k(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &final_block_size, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1302
    qmv_fast_impl
    METAL_FUNC void qmv_fast_impl(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:434
    -
    qmm_t
    void qmm_t(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &K, const constant int &N, const constant int &M, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, uint3 tid, uint lid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1310
    +
    qmm_t
    void qmm_t(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &K, const constant int &N, const constant int &M, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, uint3 tid, uint lid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1365
    adjust_matrix_offsets
    METAL_FUNC void adjust_matrix_offsets(const device T *&x, const device uint32_t *&w, const device T *&scales, const device T *&biases, device T *&y, int output_stride, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, uint3 tid)
    Definition quantized.h:1005
    -
    bs_qmv
    void bs_qmv(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &batch_ndims, const constant int *batch_shape, const device uint32_t *lhs_indices, const device uint32_t *rhs_indices, const constant size_t *lhs_strides, const constant size_t *rhs_strides, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1482
    +
    bs_qmv
    void bs_qmv(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, const constant int &batch_ndims, const constant int *batch_shape, const device uint32_t *lhs_indices, const device uint32_t *rhs_indices, const constant size_t *lhs_strides, const constant size_t *rhs_strides, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1537
    qmv_quad_impl
    METAL_FUNC void qmv_quad_impl(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, constant int &in_vec_size, const constant int &out_vec_size, uint3 tid, uint quad_gid, uint quad_lid)
    Definition quantized.h:376
    qvm
    void qvm(const device uint32_t *w, const device T *scales, const device T *biases, const device T *x, device T *y, const constant int &in_vec_size, const constant int &out_vec_size, const constant int &x_batch_ndims, const constant int *x_shape, const constant size_t *x_strides, const constant int &w_batch_ndims, const constant int *w_shape, const constant size_t *w_strides, const constant size_t *s_strides, const constant size_t *b_strides, uint3 tid, uint simd_gid, uint simd_lid)
    Definition quantized.h:1251
    qouter
    void qouter(const thread uint8_t *w, U x, U scale, U bias, thread U *result)
    Definition quantized.h:187
    diff --git a/docs/build/html/reduce__col_8h.html b/docs/build/html/reduce__col_8h.html index cd6b1d3c6..b7dda2eda 100644 --- a/docs/build/html/reduce__col_8h.html +++ b/docs/build/html/reduce__col_8h.html @@ -98,15 +98,207 @@ $(function(){ initResizable(false); });
    void qvm_split_k (const device uint32_t * w,
    const device T * scales,
    const device T * biases,
    const device T * x,
    device T * y,
    const constant int & out_vec_size,
    const constant int & x_batch_ndims,
    const constant int * x_shape,
    const constant size_t * x_strides,
    const constant int & w_batch_ndims,
    const constant int * w_shape,
    const constant size_t * w_strides,
    const constant size_t * s_strides,
    const constant size_t * b_strides,
    const constant int & final_block_size,
    - - - + + + + + + + + +

    Functions

    template<typename T , typename U , typename Op , int NDIMS, int N_READS = REDUCE_N_READS>
    void col_reduce_small (const device T *in, device U *out, const constant size_t &reduction_size, const constant size_t &reduction_stride, const constant int *shape, const constant size_t *strides, const constant int &ndim, const constant int *reduce_shape, const constant size_t *reduce_strides, const constant int &reduce_ndim, const constant size_t &non_col_reductions, uint3 gid, uint3 gsize, uint simd_lane_id, uint simd_group_id, uint3 tid, uint3 tsize)
     
    template<typename T , typename U , typename Op , int NDIMS>
    void col_reduce_small (const device T *in, device U *out, const constant size_t &reduction_size, const constant size_t &reduction_stride, const constant int *shape, const constant size_t *strides, const constant int &ndim, const constant int *reduce_shape, const constant size_t *reduce_strides, const constant int &reduce_ndim, const constant size_t &non_col_reductions, uint3 gid, uint3 gsize, uint3 lid, uint3 lsize)
     
    template<typename T , typename U , typename Op , int NDIMS>
    void col_reduce_longcolumn (const device T *in, device U *out, const constant size_t &reduction_size, const constant size_t &reduction_stride, const constant int *shape, const constant size_t *strides, const constant int &ndim, const constant int *reduce_shape, const constant size_t *reduce_strides, const constant int &reduce_ndim, const constant size_t &non_col_reductions, const constant size_t &out_size, uint3 gid, uint3 gsize, uint3 lid, uint3 lsize)
     
    template<typename T , typename U , typename Op , int NDIMS, int BM, int BN>
    void col_reduce_looped (const device T *in, device U *out, const constant size_t &reduction_size, const constant size_t &reduction_stride, const constant int *shape, const constant size_t *strides, const constant int &ndim, const constant int *reduce_shape, const constant size_t *reduce_strides, const constant int &reduce_ndim, const constant size_t &non_col_reductions, uint3 gid, uint3 gsize, uint simd_lane_id, uint simd_group_id)
     Our approach is the following simple looped approach:
     
    template<typename T , typename U , typename Op , int NDIMS, int BM, int BN>
    void col_reduce_2pass (const device T *in, device U *out, const constant size_t &reduction_size, const constant size_t &reduction_stride, const constant int *shape, const constant size_t *strides, const constant int &ndim, const constant int *reduce_shape, const constant size_t *reduce_strides, const constant int &reduce_ndim, const constant size_t &non_col_reductions, const constant size_t &out_size, uint3 gid, uint3 gsize, uint simd_lane_id, uint simd_group_id)
     

    Function Documentation

    + +

    ◆ col_reduce_2pass()

    + +
    +
    +
    +template<typename T , typename U , typename Op , int NDIMS, int BM, int BN>
    + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    void col_reduce_2pass (const device T * in,
    device U * out,
    const constant size_t & reduction_size,
    const constant size_t & reduction_stride,
    const constant int * shape,
    const constant size_t * strides,
    const constant int & ndim,
    const constant int * reduce_shape,
    const constant size_t * reduce_strides,
    const constant int & reduce_ndim,
    const constant size_t & non_col_reductions,
    const constant size_t & out_size,
    uint3 gid,
    uint3 gsize,
    uint simd_lane_id,
    uint simd_group_id )
    +
    + +
    +
    + +

    ◆ col_reduce_longcolumn()

    + +
    +
    +
    +template<typename T , typename U , typename Op , int NDIMS>
    + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
    void col_reduce_longcolumn (const device T * in,
    device U * out,
    const constant size_t & reduction_size,
    const constant size_t & reduction_stride,
    const constant int * shape,
    const constant size_t * strides,
    const constant int & ndim,
    const constant int * reduce_shape,
    const constant size_t * reduce_strides,
    const constant int & reduce_ndim,
    const constant size_t & non_col_reductions,
    const constant size_t & out_size,
    uint3 gid,
    uint3 gsize,
    uint3 lid,
    uint3 lsize )
    +
    + +
    +

    ◆ col_reduce_looped()

    @@ -204,13 +396,13 @@ template<typename T , typename U , typename Op , int NDIMS, int BM, int BN>
    - -

    ◆ col_reduce_small()

    + +

    ◆ col_reduce_small()

    -template<typename T , typename U , typename Op , int NDIMS, int N_READS = REDUCE_N_READS>
    +template<typename T , typename U , typename Op , int NDIMS>
    @@ -280,22 +472,12 @@ template<typename T , typename U , typename Op , int NDIMS, int N_READS = RED - + - - - - - - - - - - - +
    void col_reduce_small
    uint simd_lane_id, uint3 lid,
    uint simd_group_id,
    uint3 tid,
    uint3 tsize )uint3 lsize )
    diff --git a/docs/build/html/reduce__col_8h_source.html b/docs/build/html/reduce__col_8h_source.html index 835b0cd1c..caf4a5f82 100644 --- a/docs/build/html/reduce__col_8h_source.html +++ b/docs/build/html/reduce__col_8h_source.html @@ -93,334 +93,392 @@ $(function(){ initResizable(false); });
    Go to the documentation of this file.
    1// Copyright © 2023-2024 Apple Inc.
    2
    -
    3template <
    -
    4 typename T,
    -
    5 typename U,
    -
    6 typename Op,
    -
    7 int NDIMS,
    -
    8 int N_READS = REDUCE_N_READS>
    -
    -
    9[[kernel]] void col_reduce_small(
    -
    10 const device T* in [[buffer(0)]],
    -
    11 device U* out [[buffer(1)]],
    -
    12 const constant size_t& reduction_size [[buffer(2)]],
    -
    13 const constant size_t& reduction_stride [[buffer(3)]],
    -
    14 const constant int* shape [[buffer(4)]],
    -
    15 const constant size_t* strides [[buffer(5)]],
    -
    16 const constant int& ndim [[buffer(6)]],
    -
    17 const constant int* reduce_shape [[buffer(7)]],
    -
    18 const constant size_t* reduce_strides [[buffer(8)]],
    -
    19 const constant int& reduce_ndim [[buffer(9)]],
    -
    20 const constant size_t& non_col_reductions [[buffer(10)]],
    -
    21 uint3 gid [[threadgroup_position_in_grid]],
    -
    22 uint3 gsize [[threadgroups_per_grid]],
    -
    23 uint simd_lane_id [[thread_index_in_simdgroup]],
    -
    24 uint simd_group_id [[simdgroup_index_in_threadgroup]],
    -
    25 uint3 tid [[thread_position_in_grid]],
    -
    26 uint3 tsize [[threads_per_grid]]) {
    -
    27 Op op;
    -
    28 looped_elem_to_loc<NDIMS> loop;
    -
    29 const device T* row;
    -
    30
    -
    31 // Case 1: Small row small column
    -
    32 if (reduction_size * non_col_reductions < 64 && reduction_stride < 32) {
    -
    33 U totals[31];
    -
    34 for (int i = 0; i < 31; i++) {
    -
    35 totals[i] = Op::init;
    -
    36 }
    -
    37
    -
    38 short stride = reduction_stride;
    -
    39 short size = reduction_size;
    -
    40 short blocks = stride / N_READS;
    -
    41 short extra = stride - blocks * N_READS;
    -
    42
    -
    43 size_t out_idx = tid.x + tsize.y * size_t(tid.y);
    -
    44 in += elem_to_loc(out_idx, shape, strides, ndim);
    -
    45
    -
    46 for (uint r = 0; r < non_col_reductions; r++) {
    -
    47 row = in + loop.location(r, reduce_shape, reduce_strides, reduce_ndim);
    -
    48
    -
    49 for (short i = 0; i < size; i++) {
    -
    50 for (short j = 0; j < blocks; j++) {
    -
    51 for (short k = 0; k < N_READS; k++) {
    -
    52 totals[j * N_READS + k] =
    -
    53 op(totals[j * N_READS + k],
    -
    54 static_cast<U>(row[i * stride + j * N_READS + k]));
    -
    55 }
    -
    56 }
    -
    57 for (short k = 0; k < extra; k++) {
    -
    58 totals[blocks * N_READS + k] =
    -
    59 op(totals[blocks * N_READS + k],
    -
    60 static_cast<U>(row[i * stride + blocks * N_READS + k]));
    -
    61 }
    -
    62 }
    -
    63
    -
    64 loop.next(reduce_shape, reduce_strides);
    -
    65 }
    -
    66 out += out_idx * reduction_stride;
    -
    67 for (short j = 0; j < stride; j++) {
    -
    68 out[j] = totals[j];
    -
    69 }
    -
    70 }
    -
    71
    -
    72 // Case 2: Long row small column
    -
    73 else if (reduction_size * non_col_reductions < 32) {
    -
    74 U totals[N_READS];
    -
    75 for (int i = 0; i < N_READS; i++) {
    -
    76 totals[i] = Op::init;
    -
    77 }
    -
    78
    -
    79 short size = reduction_size;
    -
    80 size_t offset = size_t(tid.x) * N_READS;
    -
    81 bool safe = offset + N_READS <= reduction_stride;
    -
    82 short extra = reduction_stride - offset;
    -
    83
    -
    84 size_t out_idx = tid.y + tsize.z * size_t(tid.z);
    -
    85 in += elem_to_loc(out_idx, shape, strides, ndim) + offset;
    -
    86
    -
    87 for (uint r = 0; r < non_col_reductions; r++) {
    -
    88 row = in + loop.location(r, reduce_shape, reduce_strides, reduce_ndim);
    -
    89
    -
    90 if (safe) {
    -
    91 for (short i = 0; i < size; i++) {
    -
    92 for (short j = 0; j < N_READS; j++) {
    -
    93 totals[j] =
    -
    94 op(static_cast<U>(row[i * reduction_stride + j]), totals[j]);
    -
    95 }
    -
    96 }
    -
    97 } else {
    -
    98 for (short i = 0; i < size; i++) {
    -
    99 for (short j = 0; j < extra; j++) {
    -
    100 totals[j] =
    -
    101 op(static_cast<U>(row[i * reduction_stride + j]), totals[j]);
    -
    102 }
    -
    103 }
    -
    104 }
    -
    105
    -
    106 loop.next(reduce_shape, reduce_strides);
    -
    107 }
    -
    108 out += out_idx * reduction_stride + offset;
    -
    109 if (safe) {
    -
    110 for (short i = 0; i < N_READS; i++) {
    -
    111 out[i] = totals[i];
    -
    112 }
    -
    113 } else {
    -
    114 for (short i = 0; i < extra; i++) {
    -
    115 out[i] = totals[i];
    -
    116 }
    -
    117 }
    -
    118 }
    -
    119
    -
    120 // Case 3: Long row medium column
    -
    121 else {
    -
    122 threadgroup U shared_vals[1024];
    -
    123 U totals[N_READS];
    -
    124 for (int i = 0; i < N_READS; i++) {
    -
    125 totals[i] = Op::init;
    -
    126 }
    -
    127
    -
    128 short stride = reduction_stride;
    -
    129 short lid = simd_group_id * simd_size + simd_lane_id;
    -
    130 short2 tile((stride + N_READS - 1) / N_READS, 32);
    -
    131 short2 offset((lid % tile.x) * N_READS, lid / tile.x);
    -
    132 short sm_stride = tile.x * N_READS;
    -
    133 bool safe = offset.x + N_READS <= stride;
    -
    134
    -
    135 size_t out_idx = gid.y + gsize.y * size_t(gid.z);
    -
    136 in += elem_to_loc(out_idx, shape, strides, ndim) + offset.x;
    -
    137
    -
    138 // Read cooperatively and contiguously and aggregate the partial results.
    -
    139 size_t total = non_col_reductions * reduction_size;
    -
    140 loop.next(offset.y, reduce_shape, reduce_strides);
    -
    141 for (size_t r = offset.y; r < total; r += simd_size) {
    -
    142 row = in + loop.location(r, reduce_shape, reduce_strides, reduce_ndim);
    -
    143
    -
    144 if (safe) {
    -
    145 for (int i = 0; i < N_READS; i++) {
    -
    146 totals[i] = op(static_cast<U>(row[i]), totals[i]);
    -
    147 }
    -
    148 } else {
    -
    149 U vals[N_READS];
    -
    150 for (int i = 0; i < N_READS; i++) {
    -
    151 vals[i] = (offset.x + i < stride) ? static_cast<U>(row[i]) : op.init;
    -
    152 }
    -
    153 for (int i = 0; i < N_READS; i++) {
    -
    154 totals[i] = op(vals[i], totals[i]);
    -
    155 }
    -
    156 }
    -
    157
    -
    158 loop.next(simd_size, reduce_shape, reduce_strides);
    -
    159 }
    -
    160
    -
    161 // Each thread holds N_READS partial results but the simdgroups are not
    -
    162 // aligned to do the reduction across the simdgroup so we write our results
    -
    163 // in the shared memory and read them back according to the simdgroup.
    -
    164 for (int i = 0; i < N_READS; i++) {
    -
    165 shared_vals[offset.y * sm_stride + offset.x + i] = totals[i];
    -
    166 }
    -
    167 threadgroup_barrier(mem_flags::mem_threadgroup);
    -
    168 for (int i = 0; i < N_READS; i++) {
    -
    169 totals[i] = op.simd_reduce(
    -
    170 shared_vals[simd_lane_id * sm_stride + simd_group_id * N_READS + i]);
    -
    171 }
    -
    172
    -
    173 // Write the output.
    -
    174 if (simd_lane_id == 0) {
    -
    175 short column = simd_group_id * N_READS;
    -
    176 out += out_idx * reduction_stride + column;
    -
    177 if (column + N_READS <= stride) {
    -
    178 for (int i = 0; i < N_READS; i++) {
    -
    179 out[i] = totals[i];
    -
    180 }
    -
    181 } else {
    -
    182 for (int i = 0; column + i < stride; i++) {
    -
    183 out[i] = totals[i];
    -
    184 }
    -
    185 }
    -
    186 }
    -
    187 }
    -
    188}
    +
    3template <typename T, typename U, typename Op, int NDIMS>
    +
    +
    4[[kernel]] void col_reduce_small(
    +
    5 const device T* in [[buffer(0)]],
    +
    6 device U* out [[buffer(1)]],
    +
    7 const constant size_t& reduction_size [[buffer(2)]],
    +
    8 const constant size_t& reduction_stride [[buffer(3)]],
    +
    9 const constant int* shape [[buffer(4)]],
    +
    10 const constant size_t* strides [[buffer(5)]],
    +
    11 const constant int& ndim [[buffer(6)]],
    +
    12 const constant int* reduce_shape [[buffer(7)]],
    +
    13 const constant size_t* reduce_strides [[buffer(8)]],
    +
    14 const constant int& reduce_ndim [[buffer(9)]],
    +
    15 const constant size_t& non_col_reductions [[buffer(10)]],
    +
    16 uint3 gid [[threadgroup_position_in_grid]],
    +
    17 uint3 gsize [[threadgroups_per_grid]],
    +
    18 uint3 lid [[thread_position_in_threadgroup]],
    +
    19 uint3 lsize [[threads_per_threadgroup]]) {
    +
    20 constexpr int n_reads = 4;
    +
    21 Op op;
    +
    22 looped_elem_to_loc<NDIMS> loop;
    +
    23 const device T* row;
    +
    24
    +
    25 U totals[n_reads];
    +
    26 for (int i = 0; i < n_reads; i++) {
    +
    27 totals[i] = Op::init;
    +
    28 }
    +
    29
    +
    30 size_t column = size_t(gid.x) * lsize.x * n_reads + lid.x * n_reads;
    +
    31 if (column >= reduction_stride) {
    +
    32 return;
    +
    33 }
    +
    34 bool safe = column + n_reads <= reduction_stride;
    +
    35
    +
    36 size_t out_idx = gid.y + gsize.y * size_t(gid.z);
    +
    37 size_t in_idx = elem_to_loc(out_idx, shape, strides, ndim);
    +
    38 in += in_idx + column;
    +
    39
    +
    40 size_t total_rows = non_col_reductions * reduction_size;
    +
    41 loop.next(lid.y, reduce_shape, reduce_strides);
    +
    42 for (size_t r = lid.y; r < total_rows; r += lsize.y) {
    +
    43 row = in + loop.location(r, reduce_shape, reduce_strides, reduce_ndim);
    +
    44 if (safe) {
    +
    45 for (int i = 0; i < n_reads; i++) {
    +
    46 totals[i] = op(static_cast<U>(row[i]), totals[i]);
    +
    47 }
    +
    48 } else {
    +
    49 U vals[n_reads];
    +
    50 for (int i = 0; i < n_reads; i++) {
    +
    51 vals[i] =
    +
    52 (column + i < reduction_stride) ? static_cast<U>(row[i]) : op.init;
    +
    53 }
    +
    54 for (int i = 0; i < n_reads; i++) {
    +
    55 totals[i] = op(vals[i], totals[i]);
    +
    56 }
    +
    57 }
    +
    58 loop.next(lsize.y, reduce_shape, reduce_strides);
    +
    59 }
    +
    60
    +
    61 if (lsize.y > 1) {
    +
    62 // lsize.y should be <= 8
    +
    63 threadgroup U shared_vals[32 * 8 * n_reads];
    +
    64 for (int i = 0; i < n_reads; i++) {
    +
    65 shared_vals[lid.y * lsize.x * n_reads + lid.x * n_reads + i] = totals[i];
    +
    66 }
    +
    67 threadgroup_barrier(mem_flags::mem_threadgroup);
    +
    68 if (lid.y == 0) {
    +
    69 for (int i = 0; i < n_reads; i++) {
    +
    70 totals[i] = shared_vals[lid.x * n_reads + i];
    +
    71 }
    +
    72 for (uint j = 1; j < lsize.y; j++) {
    +
    73 for (int i = 0; i < n_reads; i++) {
    +
    74 totals[i] =
    +
    75 op(shared_vals[j * lsize.x * n_reads + lid.x * n_reads + i],
    +
    76 totals[i]);
    +
    77 }
    +
    78 }
    +
    79 }
    +
    80 }
    +
    81
    +
    82 if (lid.y == 0) {
    +
    83 out += out_idx * reduction_stride + column;
    +
    84 if (safe) {
    +
    85 for (int i = 0; i < n_reads; i++) {
    +
    86 out[i] = totals[i];
    +
    87 }
    +
    88 } else {
    +
    89 for (int i = 0; column + i < reduction_stride; i++) {
    +
    90 out[i] = totals[i];
    +
    91 }
    +
    92 }
    +
    93 }
    +
    94}
    -
    189
    -
    201template <typename T, typename U, typename Op, int NDIMS, int BM, int BN>
    -
    -
    202[[kernel]] void col_reduce_looped(
    -
    203 const device T* in [[buffer(0)]],
    -
    204 device U* out [[buffer(1)]],
    -
    205 const constant size_t& reduction_size [[buffer(2)]],
    -
    206 const constant size_t& reduction_stride [[buffer(3)]],
    -
    207 const constant int* shape [[buffer(4)]],
    -
    208 const constant size_t* strides [[buffer(5)]],
    -
    209 const constant int& ndim [[buffer(6)]],
    -
    210 const constant int* reduce_shape [[buffer(7)]],
    -
    211 const constant size_t* reduce_strides [[buffer(8)]],
    -
    212 const constant int& reduce_ndim [[buffer(9)]],
    -
    213 const constant size_t& non_col_reductions [[buffer(10)]],
    -
    214 uint3 gid [[threadgroup_position_in_grid]],
    -
    215 uint3 gsize [[threadgroups_per_grid]],
    -
    216 uint simd_lane_id [[thread_index_in_simdgroup]],
    -
    217 uint simd_group_id [[simdgroup_index_in_threadgroup]]) {
    -
    218 Op op;
    -
    219 constexpr int n_simdgroups = 4;
    -
    220 constexpr short tgp_size = n_simdgroups * simd_size;
    -
    221 constexpr short n_reads = (BM * BN) / tgp_size;
    -
    222 constexpr short n_read_blocks = BN / n_reads;
    -
    223
    -
    224 threadgroup U shared_vals[BN * BM];
    -
    225 U totals[n_reads];
    -
    226 looped_elem_to_loc<NDIMS> loop;
    -
    227 const device T* row;
    -
    228
    -
    229 for (int i = 0; i < n_reads; i++) {
    -
    230 totals[i] = Op::init;
    -
    231 }
    -
    232
    -
    233 short lid = simd_group_id * simd_size + simd_lane_id;
    -
    234 short2 offset((lid % n_read_blocks) * n_reads, lid / n_read_blocks);
    -
    235 size_t column = BN * gid.x + offset.x;
    -
    236 bool safe = column + n_reads <= reduction_stride;
    -
    237
    -
    238 size_t out_idx = gid.y + gsize.y * size_t(gid.z);
    -
    239 size_t in_idx = elem_to_loc(out_idx, shape, strides, ndim);
    -
    240 in += in_idx + column;
    -
    241
    -
    242 size_t total = non_col_reductions * reduction_size;
    -
    243 loop.next(offset.y, reduce_shape, reduce_strides);
    -
    244 for (size_t r = offset.y; r < total; r += BM) {
    -
    245 row = in + loop.location(r, reduce_shape, reduce_strides, reduce_ndim);
    -
    246
    -
    247 if (safe) {
    -
    248 for (int i = 0; i < n_reads; i++) {
    -
    249 totals[i] = op(static_cast<U>(row[i]), totals[i]);
    -
    250 }
    -
    251 } else {
    -
    252 U vals[n_reads];
    -
    253 for (int i = 0; i < n_reads; i++) {
    -
    254 vals[i] =
    -
    255 (column + i < reduction_stride) ? static_cast<U>(row[i]) : op.init;
    -
    256 }
    -
    257 for (int i = 0; i < n_reads; i++) {
    -
    258 totals[i] = op(vals[i], totals[i]);
    -
    259 }
    -
    260 }
    -
    261
    -
    262 loop.next(BM, reduce_shape, reduce_strides);
    -
    263 }
    -
    264
    -
    265 // We can use a simd reduction to accumulate across BM so each thread writes
    -
    266 // the partial output to SM and then each simdgroup does BN / n_simdgroups
    -
    267 // accumulations.
    -
    268 if (BM == 32) {
    -
    269 constexpr int n_outputs = BN / n_simdgroups;
    -
    270 static_assert(
    -
    271 BM != 32 || n_outputs == n_reads,
    -
    272 "The tile should be selected such that n_outputs == n_reads");
    -
    273 for (int i = 0; i < n_reads; i++) {
    -
    274 shared_vals[offset.y * BN + offset.x + i] = totals[i];
    -
    275 }
    -
    276 threadgroup_barrier(mem_flags::mem_threadgroup);
    -
    277 short2 out_offset(simd_group_id * n_outputs, simd_lane_id);
    -
    278 for (int i = 0; i < n_outputs; i++) {
    -
    279 totals[i] =
    -
    280 op.simd_reduce(shared_vals[out_offset.y * BN + out_offset.x + i]);
    -
    281 }
    -
    282
    -
    283 // Write the output.
    -
    284 if (simd_lane_id == 0) {
    -
    285 size_t out_column = BN * gid.x + out_offset.x;
    -
    286 out += out_idx * reduction_stride + out_column;
    -
    287 if (out_column + n_outputs <= reduction_stride) {
    -
    288 for (int i = 0; i < n_outputs; i++) {
    -
    289 out[i] = totals[i];
    -
    290 }
    -
    291 } else {
    -
    292 for (int i = 0; out_column + i < reduction_stride; i++) {
    -
    293 out[i] = totals[i];
    -
    294 }
    -
    295 }
    -
    296 }
    -
    297 }
    -
    298
    -
    299 // Each thread holds n_reads partial results. We write them all out to shared
    -
    300 // memory and threads with offset.y == 0 aggregate the columns and write the
    -
    301 // outputs.
    -
    302 else {
    -
    303 short x_block = offset.x / n_reads;
    -
    304 for (int i = 0; i < n_reads; i++) {
    -
    305 shared_vals[x_block * BM * n_reads + i * BM + offset.y] = totals[i];
    -
    306 }
    -
    307 threadgroup_barrier(mem_flags::mem_threadgroup);
    -
    308 if (offset.y == 0) {
    -
    309 for (int i = 0; i < n_reads; i++) {
    -
    310 for (int j = 1; j < BM; j++) {
    -
    311 totals[i] =
    -
    312 op(shared_vals[x_block * BM * n_reads + i * BM + j], totals[i]);
    -
    313 }
    -
    314 }
    -
    315 }
    -
    316
    -
    317 // Write the output.
    -
    318 if (offset.y == 0) {
    -
    319 out += out_idx * reduction_stride + column;
    -
    320 if (safe) {
    -
    321 for (int i = 0; i < n_reads; i++) {
    -
    322 out[i] = totals[i];
    -
    323 }
    -
    324 } else {
    -
    325 for (int i = 0; column + i < reduction_stride; i++) {
    -
    326 out[i] = totals[i];
    -
    327 }
    -
    328 }
    -
    329 }
    -
    330 }
    -
    331}
    +
    95
    +
    96template <typename T, typename U, typename Op, int NDIMS>
    +
    +
    97[[kernel]] void col_reduce_longcolumn(
    +
    98 const device T* in [[buffer(0)]],
    +
    99 device U* out [[buffer(1)]],
    +
    100 const constant size_t& reduction_size [[buffer(2)]],
    +
    101 const constant size_t& reduction_stride [[buffer(3)]],
    +
    102 const constant int* shape [[buffer(4)]],
    +
    103 const constant size_t* strides [[buffer(5)]],
    +
    104 const constant int& ndim [[buffer(6)]],
    +
    105 const constant int* reduce_shape [[buffer(7)]],
    +
    106 const constant size_t* reduce_strides [[buffer(8)]],
    +
    107 const constant int& reduce_ndim [[buffer(9)]],
    +
    108 const constant size_t& non_col_reductions [[buffer(10)]],
    +
    109 const constant size_t& out_size [[buffer(11)]],
    +
    110 uint3 gid [[threadgroup_position_in_grid]],
    +
    111 uint3 gsize [[threadgroups_per_grid]],
    +
    112 uint3 lid [[thread_position_in_threadgroup]],
    +
    113 uint3 lsize [[threads_per_threadgroup]]) {
    +
    114 Op op;
    +
    115 looped_elem_to_loc<NDIMS> loop;
    +
    116 const device T* row;
    +
    117
    +
    118 size_t out_idx = gid.x + gsize.x * size_t(gid.y);
    +
    119 size_t in_idx = elem_to_loc(out_idx, shape, strides, ndim);
    +
    120 in += in_idx + lid.x;
    +
    121
    +
    122 U total = Op::init;
    +
    123 size_t total_rows = non_col_reductions * reduction_size;
    +
    124 loop.next(gid.z * lsize.y + lid.y, reduce_shape, reduce_strides);
    +
    125 for (size_t r = gid.z * lsize.y + lid.y; r < total_rows;
    +
    126 r += lsize.y * gsize.z) {
    +
    127 row = in + loop.location(r, reduce_shape, reduce_strides, reduce_ndim);
    +
    128 total = op(static_cast<U>(*row), total);
    +
    129 loop.next(lsize.y * gsize.z, reduce_shape, reduce_strides);
    +
    130 }
    +
    131
    +
    132 threadgroup U shared_vals[32 * 32];
    +
    133 shared_vals[lid.y * lsize.x + lid.x] = total;
    +
    134 threadgroup_barrier(mem_flags::mem_threadgroup);
    +
    135 if (lid.y == 0) {
    +
    136 for (uint i = 1; i < lsize.y; i++) {
    +
    137 total = op(total, shared_vals[i * lsize.x + lid.x]);
    +
    138 }
    +
    139 out[gid.z * out_size + out_idx * reduction_stride + lid.x] = total;
    +
    140 }
    +
    141}
    +
    +
    142
    +
    154template <typename T, typename U, typename Op, int NDIMS, int BM, int BN>
    +
    +
    155[[kernel]] void col_reduce_looped(
    +
    156 const device T* in [[buffer(0)]],
    +
    157 device U* out [[buffer(1)]],
    +
    158 const constant size_t& reduction_size [[buffer(2)]],
    +
    159 const constant size_t& reduction_stride [[buffer(3)]],
    +
    160 const constant int* shape [[buffer(4)]],
    +
    161 const constant size_t* strides [[buffer(5)]],
    +
    162 const constant int& ndim [[buffer(6)]],
    +
    163 const constant int* reduce_shape [[buffer(7)]],
    +
    164 const constant size_t* reduce_strides [[buffer(8)]],
    +
    165 const constant int& reduce_ndim [[buffer(9)]],
    +
    166 const constant size_t& non_col_reductions [[buffer(10)]],
    +
    167 uint3 gid [[threadgroup_position_in_grid]],
    +
    168 uint3 gsize [[threadgroups_per_grid]],
    +
    169 uint simd_lane_id [[thread_index_in_simdgroup]],
    +
    170 uint simd_group_id [[simdgroup_index_in_threadgroup]]) {
    +
    171 Op op;
    +
    172 constexpr int n_simdgroups = 8;
    +
    173 constexpr short tgp_size = n_simdgroups * simd_size;
    +
    174 constexpr short n_reads = (BM * BN) / tgp_size;
    +
    175 constexpr short n_read_blocks = BN / n_reads;
    +
    176
    +
    177 threadgroup U shared_vals[BN * BM];
    +
    178 U totals[n_reads];
    +
    179 looped_elem_to_loc<NDIMS> loop;
    +
    180 const device T* row;
    +
    181
    +
    182 for (int i = 0; i < n_reads; i++) {
    +
    183 totals[i] = Op::init;
    +
    184 }
    +
    185
    +
    186 short lid = simd_group_id * simd_size + simd_lane_id;
    +
    187 short2 offset((lid % n_read_blocks) * n_reads, lid / n_read_blocks);
    +
    188 size_t column = BN * gid.x + offset.x;
    +
    189 bool safe = column + n_reads <= reduction_stride;
    +
    190
    +
    191 size_t out_idx = gid.y + gsize.y * size_t(gid.z);
    +
    192 size_t in_idx = elem_to_loc(out_idx, shape, strides, ndim);
    +
    193 in += in_idx + column;
    +
    194
    +
    195 size_t total = non_col_reductions * reduction_size;
    +
    196 loop.next(offset.y, reduce_shape, reduce_strides);
    +
    197 for (size_t r = offset.y; r < total; r += BM) {
    +
    198 row = in + loop.location(r, reduce_shape, reduce_strides, reduce_ndim);
    +
    199
    +
    200 if (safe) {
    +
    201 for (int i = 0; i < n_reads; i++) {
    +
    202 totals[i] = op(static_cast<U>(row[i]), totals[i]);
    +
    203 }
    +
    204 } else {
    +
    205 U vals[n_reads];
    +
    206 for (int i = 0; i < n_reads; i++) {
    +
    207 vals[i] =
    +
    208 (column + i < reduction_stride) ? static_cast<U>(row[i]) : op.init;
    +
    209 }
    +
    210 for (int i = 0; i < n_reads; i++) {
    +
    211 totals[i] = op(vals[i], totals[i]);
    +
    212 }
    +
    213 }
    +
    214
    +
    215 loop.next(BM, reduce_shape, reduce_strides);
    +
    216 }
    +
    217
    +
    218 // We can use a simd reduction to accumulate across BM so each thread writes
    +
    219 // the partial output to SM and then each simdgroup does BN / n_simdgroups
    +
    220 // accumulations.
    +
    221 if (BM == 32) {
    +
    222 constexpr int n_outputs = BN / n_simdgroups;
    +
    223 static_assert(
    +
    224 BM != 32 || n_outputs == n_reads,
    +
    225 "The tile should be selected such that n_outputs == n_reads");
    +
    226 for (int i = 0; i < n_reads; i++) {
    +
    227 shared_vals[offset.y * BN + offset.x + i] = totals[i];
    +
    228 }
    +
    229 threadgroup_barrier(mem_flags::mem_threadgroup);
    +
    230 short2 out_offset(simd_group_id * n_outputs, simd_lane_id);
    +
    231 for (int i = 0; i < n_outputs; i++) {
    +
    232 totals[i] =
    +
    233 op.simd_reduce(shared_vals[out_offset.y * BN + out_offset.x + i]);
    +
    234 }
    +
    235
    +
    236 // Write the output.
    +
    237 if (simd_lane_id == 0) {
    +
    238 size_t out_column = BN * gid.x + out_offset.x;
    +
    239 out += out_idx * reduction_stride + out_column;
    +
    240 if (out_column + n_outputs <= reduction_stride) {
    +
    241 for (int i = 0; i < n_outputs; i++) {
    +
    242 out[i] = totals[i];
    +
    243 }
    +
    244 } else {
    +
    245 for (int i = 0; out_column + i < reduction_stride; i++) {
    +
    246 out[i] = totals[i];
    +
    247 }
    +
    248 }
    +
    249 }
    +
    250 }
    +
    251
    +
    252 // Each thread holds n_reads partial results. We write them all out to shared
    +
    253 // memory and threads with offset.y == 0 aggregate the columns and write the
    +
    254 // outputs.
    +
    255 else {
    +
    256 short x_block = offset.x / n_reads;
    +
    257 for (int i = 0; i < n_reads; i++) {
    +
    258 shared_vals[x_block * BM * n_reads + i * BM + offset.y] = totals[i];
    +
    259 }
    +
    260 threadgroup_barrier(mem_flags::mem_threadgroup);
    +
    261 if (offset.y == 0) {
    +
    262 for (int i = 0; i < n_reads; i++) {
    +
    263 for (int j = 1; j < BM; j++) {
    +
    264 totals[i] =
    +
    265 op(shared_vals[x_block * BM * n_reads + i * BM + j], totals[i]);
    +
    266 }
    +
    267 }
    +
    268 }
    +
    269
    +
    270 // Write the output.
    +
    271 if (offset.y == 0) {
    +
    272 out += out_idx * reduction_stride + column;
    +
    273 if (safe) {
    +
    274 for (int i = 0; i < n_reads; i++) {
    +
    275 out[i] = totals[i];
    +
    276 }
    +
    277 } else {
    +
    278 for (int i = 0; column + i < reduction_stride; i++) {
    +
    279 out[i] = totals[i];
    +
    280 }
    +
    281 }
    +
    282 }
    +
    283 }
    +
    284}
    +
    +
    285
    +
    286template <typename T, typename U, typename Op, int NDIMS, int BM, int BN>
    +
    +
    287[[kernel]] void col_reduce_2pass(
    +
    288 const device T* in [[buffer(0)]],
    +
    289 device U* out [[buffer(1)]],
    +
    290 const constant size_t& reduction_size [[buffer(2)]],
    +
    291 const constant size_t& reduction_stride [[buffer(3)]],
    +
    292 const constant int* shape [[buffer(4)]],
    +
    293 const constant size_t* strides [[buffer(5)]],
    +
    294 const constant int& ndim [[buffer(6)]],
    +
    295 const constant int* reduce_shape [[buffer(7)]],
    +
    296 const constant size_t* reduce_strides [[buffer(8)]],
    +
    297 const constant int& reduce_ndim [[buffer(9)]],
    +
    298 const constant size_t& non_col_reductions [[buffer(10)]],
    +
    299 const constant size_t& out_size [[buffer(11)]],
    +
    300 uint3 gid [[threadgroup_position_in_grid]],
    +
    301 uint3 gsize [[threadgroups_per_grid]],
    +
    302 uint simd_lane_id [[thread_index_in_simdgroup]],
    +
    303 uint simd_group_id [[simdgroup_index_in_threadgroup]]) {
    +
    304 Op op;
    +
    305 constexpr int n_simdgroups = 8;
    +
    306 constexpr short tgp_size = n_simdgroups * simd_size;
    +
    307 constexpr short n_reads = (BM * BN) / tgp_size;
    +
    308 constexpr short n_read_blocks = BN / n_reads;
    +
    309 constexpr int n_outputs = BN / n_simdgroups;
    +
    310 constexpr short outer_blocks = 32;
    +
    311 static_assert(BM == 32, "BM should be equal to 32");
    +
    312
    +
    313 threadgroup U shared_vals[BN * BM];
    +
    314 U totals[n_reads];
    +
    315 looped_elem_to_loc<NDIMS> loop;
    +
    316 const device T* row;
    +
    317
    +
    318 for (int i = 0; i < n_reads; i++) {
    +
    319 totals[i] = Op::init;
    +
    320 }
    +
    321
    +
    322 short lid = simd_group_id * simd_size + simd_lane_id;
    +
    323 short2 offset((lid % n_read_blocks) * n_reads, lid / n_read_blocks);
    +
    324 size_t column = BN * gid.x + offset.x;
    +
    325 bool safe = column + n_reads <= reduction_stride;
    +
    326
    +
    327 size_t full_idx = gid.y + gsize.y * size_t(gid.z);
    +
    328 size_t block_idx = full_idx / out_size;
    +
    329 size_t out_idx = full_idx % out_size;
    +
    330 size_t in_idx = elem_to_loc(out_idx, shape, strides, ndim);
    +
    331 in += in_idx + column;
    +
    332
    +
    333 size_t total = non_col_reductions * reduction_size;
    +
    334 loop.next(offset.y + block_idx * BM, reduce_shape, reduce_strides);
    +
    335 for (size_t r = offset.y + block_idx * BM; r < total;
    +
    336 r += outer_blocks * BM) {
    +
    337 row = in + loop.location(r, reduce_shape, reduce_strides, reduce_ndim);
    +
    338
    +
    339 if (safe) {
    +
    340 for (int i = 0; i < n_reads; i++) {
    +
    341 totals[i] = op(static_cast<U>(row[i]), totals[i]);
    +
    342 }
    +
    343 } else {
    +
    344 U vals[n_reads];
    +
    345 for (int i = 0; i < n_reads; i++) {
    +
    346 vals[i] =
    +
    347 (column + i < reduction_stride) ? static_cast<U>(row[i]) : op.init;
    +
    348 }
    +
    349 for (int i = 0; i < n_reads; i++) {
    +
    350 totals[i] = op(vals[i], totals[i]);
    +
    351 }
    +
    352 }
    +
    353
    +
    354 loop.next(outer_blocks * BM, reduce_shape, reduce_strides);
    +
    355 }
    +
    356
    +
    357 // We can use a simd reduction to accumulate across BM so each thread writes
    +
    358 // the partial output to SM and then each simdgroup does BN / n_simdgroups
    +
    359 // accumulations.
    +
    360 for (int i = 0; i < n_reads; i++) {
    +
    361 shared_vals[offset.y * BN + offset.x + i] = totals[i];
    +
    362 }
    +
    363 threadgroup_barrier(mem_flags::mem_threadgroup);
    +
    364 short2 out_offset(simd_group_id * n_outputs, simd_lane_id);
    +
    365 for (int i = 0; i < n_outputs; i++) {
    +
    366 totals[i] =
    +
    367 op.simd_reduce(shared_vals[out_offset.y * BN + out_offset.x + i]);
    +
    368 }
    +
    369
    +
    370 // Write the output.
    +
    371 if (simd_lane_id == 0) {
    +
    372 size_t out_column = BN * gid.x + out_offset.x;
    +
    373 out += full_idx * reduction_stride + out_column;
    +
    374 if (out_column + n_outputs <= reduction_stride) {
    +
    375 for (int i = 0; i < n_outputs; i++) {
    +
    376 out[i] = totals[i];
    +
    377 }
    +
    378 } else {
    +
    379 for (int i = 0; out_column + i < reduction_stride; i++) {
    +
    380 out[i] = totals[i];
    +
    381 }
    +
    382 }
    +
    383 }
    +
    384}
    simd_size
    static constant constexpr const uint8_t simd_size
    Definition ops.h:22
    elem_to_loc
    METAL_FUNC stride_t elem_to_loc(uint elem, constant const int *shape, constant const stride_t *strides, int ndim)
    Definition utils.h:87
    op
    Op op
    Definition binary.h:129
    -
    REDUCE_N_READS
    static constexpr int REDUCE_N_READS
    Definition defines.h:12
    -
    col_reduce_looped
    void col_reduce_looped(const device T *in, device U *out, const constant size_t &reduction_size, const constant size_t &reduction_stride, const constant int *shape, const constant size_t *strides, const constant int &ndim, const constant int *reduce_shape, const constant size_t *reduce_strides, const constant int &reduce_ndim, const constant size_t &non_col_reductions, uint3 gid, uint3 gsize, uint simd_lane_id, uint simd_group_id)
    Our approach is the following simple looped approach:
    Definition reduce_col.h:202
    -
    col_reduce_small
    void col_reduce_small(const device T *in, device U *out, const constant size_t &reduction_size, const constant size_t &reduction_stride, const constant int *shape, const constant size_t *strides, const constant int &ndim, const constant int *reduce_shape, const constant size_t *reduce_strides, const constant int &reduce_ndim, const constant size_t &non_col_reductions, uint3 gid, uint3 gsize, uint simd_lane_id, uint simd_group_id, uint3 tid, uint3 tsize)
    Definition reduce_col.h:9
    +
    col_reduce_2pass
    void col_reduce_2pass(const device T *in, device U *out, const constant size_t &reduction_size, const constant size_t &reduction_stride, const constant int *shape, const constant size_t *strides, const constant int &ndim, const constant int *reduce_shape, const constant size_t *reduce_strides, const constant int &reduce_ndim, const constant size_t &non_col_reductions, const constant size_t &out_size, uint3 gid, uint3 gsize, uint simd_lane_id, uint simd_group_id)
    Definition reduce_col.h:287
    +
    col_reduce_looped
    void col_reduce_looped(const device T *in, device U *out, const constant size_t &reduction_size, const constant size_t &reduction_stride, const constant int *shape, const constant size_t *strides, const constant int &ndim, const constant int *reduce_shape, const constant size_t *reduce_strides, const constant int &reduce_ndim, const constant size_t &non_col_reductions, uint3 gid, uint3 gsize, uint simd_lane_id, uint simd_group_id)
    Our approach is the following simple looped approach:
    Definition reduce_col.h:155
    +
    col_reduce_longcolumn
    void col_reduce_longcolumn(const device T *in, device U *out, const constant size_t &reduction_size, const constant size_t &reduction_stride, const constant int *shape, const constant size_t *strides, const constant int &ndim, const constant int *reduce_shape, const constant size_t *reduce_strides, const constant int &reduce_ndim, const constant size_t &non_col_reductions, const constant size_t &out_size, uint3 gid, uint3 gsize, uint3 lid, uint3 lsize)
    Definition reduce_col.h:97
    +
    col_reduce_small
    void col_reduce_small(const device T *in, device U *out, const constant size_t &reduction_size, const constant size_t &reduction_stride, const constant int *shape, const constant size_t *strides, const constant int &ndim, const constant int *reduce_shape, const constant size_t *reduce_strides, const constant int &reduce_ndim, const constant size_t &non_col_reductions, uint3 gid, uint3 gsize, uint3 lid, uint3 lsize)
    Definition reduce_col.h:4
    looped_elem_to_loc
    Definition utils.h:197
    looped_elem_to_loc::next
    void next(const constant int *shape, const constant size_t *strides)
    Definition utils.h:202
    looped_elem_to_loc::location
    offset_t location(offset_t, const constant int *, const constant size_t *, int)
    Definition utils.h:229
    diff --git a/docs/build/html/sdpa__vector_8h.html b/docs/build/html/sdpa__vector_8h.html index df9d5fd42..daee8ad0a 100644 --- a/docs/build/html/sdpa__vector_8h.html +++ b/docs/build/html/sdpa__vector_8h.html @@ -99,13 +99,13 @@ $(function(){ initResizable(false); }); - - - + + +

    Functions

    template<typename T , int D>
    void sdpa_vector (const device T *queries, const device T *keys, const device T *values, device T *out, const constant int &gqa_factor, const constant int &N, const constant size_t &k_stride, const constant float &scale, uint3 tid, uint simd_gid, uint simd_lid)
     
    template<typename T , int D>
    void sdpa_vector (const device T *queries, const device T *keys, const device T *values, device T *out, const constant int &gqa_factor, const constant int &N, const constant size_t &k_stride, const constant size_t &v_stride, const constant float &scale, uint3 tid, uint simd_gid, uint simd_lid)
     

    Function Documentation

    - -

    ◆ sdpa_vector()

    + +

    ◆ sdpa_vector()

    @@ -147,6 +147,11 @@ template<typename T , int D>
    const constant size_t & k_stride, + + + + const constant size_t & v_stride, + diff --git a/docs/build/html/sdpa__vector_8h_source.html b/docs/build/html/sdpa__vector_8h_source.html index 2d695b09b..390193135 100644 --- a/docs/build/html/sdpa__vector_8h_source.html +++ b/docs/build/html/sdpa__vector_8h_source.html @@ -99,7 +99,7 @@ $(function(){ initResizable(false); });
    6
    7template <typename T, int D>
    -
    8[[kernel]] void sdpa_vector(
    +
    8[[kernel]] void sdpa_vector(
    9 const device T* queries [[buffer(0)]],
    10 const device T* keys [[buffer(1)]],
    11 const device T* values [[buffer(2)]],
    @@ -107,113 +107,114 @@ $(function(){ initResizable(false); });
    13 const constant int& gqa_factor,
    14 const constant int& N,
    15 const constant size_t& k_stride,
    -
    16 const constant float& scale,
    -
    17 uint3 tid [[threadgroup_position_in_grid]],
    -
    18 uint simd_gid [[simdgroup_index_in_threadgroup]],
    -
    19 uint simd_lid [[thread_index_in_simdgroup]]) {
    -
    20 constexpr int BN = 32;
    -
    21 constexpr int BD = 32;
    -
    22 constexpr int elem_per_thread = D / BD;
    -
    23
    -
    24 const int stride = BN * D;
    -
    25
    -
    26 typedef float U;
    -
    27
    -
    28 thread U q[elem_per_thread];
    -
    29 thread U k[elem_per_thread];
    -
    30 thread U o[elem_per_thread];
    -
    31
    -
    32 threadgroup U outputs[BN * BD];
    -
    33 threadgroup U max_scores[BN];
    -
    34 threadgroup U sum_exp_scores[BN];
    -
    35
    -
    36 // Adjust positions
    -
    37 const int head_idx = tid.y;
    -
    38 const int kv_head_idx = head_idx / gqa_factor;
    -
    39 queries += head_idx * D + simd_lid * elem_per_thread;
    -
    40 keys += kv_head_idx * k_stride + simd_gid * D + simd_lid * elem_per_thread;
    -
    41 values += kv_head_idx * k_stride + simd_gid * D + simd_lid * elem_per_thread;
    -
    42 out += head_idx * D + simd_gid * elem_per_thread;
    -
    43
    -
    44 // Read the query and 0 the output accumulator
    -
    45 for (int i = 0; i < elem_per_thread; i++) {
    -
    46 q[i] = static_cast<U>(scale) * queries[i];
    -
    47 }
    -
    48 for (int i = 0; i < elem_per_thread; i++) {
    -
    49 o[i] = 0;
    -
    50 }
    -
    51
    -
    52 U max_score = -INFINITY;
    -
    53 U sum_exp_score = 0;
    -
    54
    -
    55 // For each key
    -
    56 for (int i = simd_gid; i < N; i += BN) {
    -
    57 // Read the key
    -
    58 for (int i = 0; i < elem_per_thread; i++) {
    -
    59 k[i] = keys[i];
    -
    60 }
    -
    61
    -
    62 // Compute the i-th score
    -
    63 U score = 0;
    -
    64 for (int i = 0; i < elem_per_thread; i++) {
    -
    65 score += q[i] * k[i];
    -
    66 }
    -
    67 score = simd_sum(score);
    -
    68
    -
    69 // Update the accumulators
    -
    70 U new_max = max(max_score, score);
    -
    71 U factor = fast::exp(max_score - new_max);
    -
    72 U exp_score = fast::exp(score - new_max);
    -
    73
    -
    74 max_score = new_max;
    -
    75 sum_exp_score = sum_exp_score * factor + exp_score;
    -
    76
    -
    77 // Update the output accumulator
    -
    78 for (int i = 0; i < elem_per_thread; i++) {
    -
    79 o[i] = o[i] * factor + exp_score * values[i];
    -
    80 }
    -
    81
    -
    82 // Move the pointers to the next kv
    -
    83 keys += stride;
    -
    84 values += stride;
    -
    85 }
    -
    86 threadgroup_barrier(mem_flags::mem_threadgroup);
    -
    87
    -
    88 // Each thread has a partial part of the output so we need to combine them.
    -
    89
    -
    90 // First let's communicate the max and sum_exp
    -
    91 if (simd_lid == 0) {
    -
    92 max_scores[simd_gid] = max_score;
    -
    93 sum_exp_scores[simd_gid] = sum_exp_score;
    -
    94 }
    -
    95 threadgroup_barrier(mem_flags::mem_threadgroup);
    -
    96 max_score = max_scores[simd_lid];
    -
    97 U new_max = simd_max(max_score);
    -
    98 U factor = fast::exp(max_score - new_max);
    -
    99 sum_exp_score = simd_sum(sum_exp_scores[simd_lid] * factor);
    -
    100
    -
    101 // Now we need to aggregate all the outputs
    -
    102 for (int i = 0; i < elem_per_thread; i++) {
    -
    103 outputs[simd_lid * BD + simd_gid] = o[i];
    -
    104 threadgroup_barrier(mem_flags::mem_threadgroup);
    -
    105 o[i] = simd_sum(outputs[simd_gid * BD + simd_lid] * factor) / sum_exp_score;
    -
    106 threadgroup_barrier(mem_flags::mem_threadgroup);
    -
    107 }
    -
    108
    -
    109 // And write the output
    -
    110 if (simd_lid == 0) {
    -
    111 for (int i = 0; i < elem_per_thread; i++) {
    -
    112 out[i] = static_cast<T>(o[i]);
    -
    113 }
    -
    114 }
    -
    115}
    +
    16 const constant size_t& v_stride,
    +
    17 const constant float& scale,
    +
    18 uint3 tid [[threadgroup_position_in_grid]],
    +
    19 uint simd_gid [[simdgroup_index_in_threadgroup]],
    +
    20 uint simd_lid [[thread_index_in_simdgroup]]) {
    +
    21 constexpr int BN = 32;
    +
    22 constexpr int BD = 32;
    +
    23 constexpr int elem_per_thread = D / BD;
    +
    24
    +
    25 const int stride = BN * D;
    +
    26
    +
    27 typedef float U;
    +
    28
    +
    29 thread U q[elem_per_thread];
    +
    30 thread U k[elem_per_thread];
    +
    31 thread U o[elem_per_thread];
    +
    32
    +
    33 threadgroup U outputs[BN * BD];
    +
    34 threadgroup U max_scores[BN];
    +
    35 threadgroup U sum_exp_scores[BN];
    +
    36
    +
    37 // Adjust positions
    +
    38 const int head_idx = tid.y;
    +
    39 const int kv_head_idx = head_idx / gqa_factor;
    +
    40 queries += head_idx * D + simd_lid * elem_per_thread;
    +
    41 keys += kv_head_idx * k_stride + simd_gid * D + simd_lid * elem_per_thread;
    +
    42 values += kv_head_idx * v_stride + simd_gid * D + simd_lid * elem_per_thread;
    +
    43 out += head_idx * D + simd_gid * elem_per_thread;
    +
    44
    +
    45 // Read the query and 0 the output accumulator
    +
    46 for (int i = 0; i < elem_per_thread; i++) {
    +
    47 q[i] = static_cast<U>(scale) * queries[i];
    +
    48 }
    +
    49 for (int i = 0; i < elem_per_thread; i++) {
    +
    50 o[i] = 0;
    +
    51 }
    +
    52
    +
    53 U max_score = -INFINITY;
    +
    54 U sum_exp_score = 0;
    +
    55
    +
    56 // For each key
    +
    57 for (int i = simd_gid; i < N; i += BN) {
    +
    58 // Read the key
    +
    59 for (int i = 0; i < elem_per_thread; i++) {
    +
    60 k[i] = keys[i];
    +
    61 }
    +
    62
    +
    63 // Compute the i-th score
    +
    64 U score = 0;
    +
    65 for (int i = 0; i < elem_per_thread; i++) {
    +
    66 score += q[i] * k[i];
    +
    67 }
    +
    68 score = simd_sum(score);
    +
    69
    +
    70 // Update the accumulators
    +
    71 U new_max = max(max_score, score);
    +
    72 U factor = fast::exp(max_score - new_max);
    +
    73 U exp_score = fast::exp(score - new_max);
    +
    74
    +
    75 max_score = new_max;
    +
    76 sum_exp_score = sum_exp_score * factor + exp_score;
    +
    77
    +
    78 // Update the output accumulator
    +
    79 for (int i = 0; i < elem_per_thread; i++) {
    +
    80 o[i] = o[i] * factor + exp_score * values[i];
    +
    81 }
    +
    82
    +
    83 // Move the pointers to the next kv
    +
    84 keys += stride;
    +
    85 values += stride;
    +
    86 }
    +
    87 threadgroup_barrier(mem_flags::mem_threadgroup);
    +
    88
    +
    89 // Each thread has a partial part of the output so we need to combine them.
    +
    90
    +
    91 // First let's communicate the max and sum_exp
    +
    92 if (simd_lid == 0) {
    +
    93 max_scores[simd_gid] = max_score;
    +
    94 sum_exp_scores[simd_gid] = sum_exp_score;
    +
    95 }
    +
    96 threadgroup_barrier(mem_flags::mem_threadgroup);
    +
    97 max_score = max_scores[simd_lid];
    +
    98 U new_max = simd_max(max_score);
    +
    99 U factor = fast::exp(max_score - new_max);
    +
    100 sum_exp_score = simd_sum(sum_exp_scores[simd_lid] * factor);
    +
    101
    +
    102 // Now we need to aggregate all the outputs
    +
    103 for (int i = 0; i < elem_per_thread; i++) {
    +
    104 outputs[simd_lid * BD + simd_gid] = o[i];
    +
    105 threadgroup_barrier(mem_flags::mem_threadgroup);
    +
    106 o[i] = simd_sum(outputs[simd_gid * BD + simd_lid] * factor) / sum_exp_score;
    +
    107 threadgroup_barrier(mem_flags::mem_threadgroup);
    +
    108 }
    +
    109
    +
    110 // And write the output
    +
    111 if (simd_lid == 0) {
    +
    112 for (int i = 0; i < elem_per_thread; i++) {
    +
    113 out[i] = static_cast<T>(o[i]);
    +
    114 }
    +
    115 }
    +
    116}
    metal::fast::exp
    METAL_FUNC bfloat16_t exp(bfloat16_t x)
    Definition bf16_math.h:242
    metal
    Definition bf16.h:265
    metal::simd_max
    METAL_FUNC bfloat16_t simd_max(bfloat16_t data)
    Definition bf16_math.h:392
    metal::simd_sum
    METAL_FUNC bfloat16_t simd_sum(bfloat16_t data)
    Definition bf16_math.h:392
    metal::max
    METAL_FUNC bfloat16_t max(bfloat16_t x, bfloat16_t y)
    Definition bf16_math.h:234
    -
    sdpa_vector
    void sdpa_vector(const device T *queries, const device T *keys, const device T *values, device T *out, const constant int &gqa_factor, const constant int &N, const constant size_t &k_stride, const constant float &scale, uint3 tid, uint simd_gid, uint simd_lid)
    Definition sdpa_vector.h:8
    +
    sdpa_vector
    void sdpa_vector(const device T *queries, const device T *keys, const device T *values, device T *out, const constant int &gqa_factor, const constant int &N, const constant size_t &k_stride, const constant size_t &v_stride, const constant float &scale, uint3 tid, uint simd_gid, uint simd_lid)
    Definition sdpa_vector.h:8
    diff --git a/docs/build/html/search/all_1.js b/docs/build/html/search/all_1.js index 1ef7de0c9..2feceb6ce 100644 --- a/docs/build/html/search/all_1.js +++ b/docs/build/html/search/all_1.js @@ -38,7 +38,7 @@ var searchData= ['all_35',['all',['../group__ops.html#ga3b1b90ef1275ca17655b6d7f25d3ee68',1,'mlx::core::all(const array &a, bool keepdims, StreamOrDevice s={})'],['../group__ops.html#ga3689e12e8f42dadb4cbe2b07dc4099f4',1,'mlx::core::all(const array &a, StreamOrDevice s={})'],['../group__ops.html#gac0919c6ba53aea35a7683dea7e9a9a59',1,'mlx::core::all(const array &a, const std::vector< int > &axes, bool keepdims=false, StreamOrDevice s={})'],['../group__ops.html#gae2d5fcc5b62d673cca76c08b7b4afbbc',1,'mlx::core::all(const array &a, int axis, bool keepdims=false, StreamOrDevice s={})']]], ['all_5fgather_36',['all_gather',['../namespacemlx_1_1core_1_1distributed_1_1detail.html#aeb5a1726358213bc75756506f7b54d04',1,'mlx::core::distributed::detail::all_gather()'],['../namespacemlx_1_1core_1_1distributed.html#a82ef5e8cc7ac62cd228e51b1c1b77cb7',1,'mlx::core::distributed::all_gather()']]], ['all_5freduce_37',['all_reduce',['../reduce__all_8h.html#a99ef48ae72b3e715c5f4d7ea07cd213d',1,'reduce_all.h']]], - ['all_5freduce_5fdispatch_38',['all_reduce_dispatch',['../namespacemlx_1_1core.html#af7b7ca7c6aa87558d9f98cee5c7a99a8',1,'mlx::core']]], + ['all_5freduce_5fdispatch_38',['all_reduce_dispatch',['../namespacemlx_1_1core.html#a3ab0fd997d9a35782106ff083a72e098',1,'mlx::core']]], ['all_5fsum_39',['all_sum',['../namespacemlx_1_1core_1_1distributed_1_1detail.html#aa1d225b25f7b6426c48c5e35860ee960',1,'mlx::core::distributed::detail::all_sum()'],['../namespacemlx_1_1core_1_1distributed.html#a67ccb1a5445fc6f5db49dd36a15e5980',1,'mlx::core::distributed::all_sum()']]], ['allclose_40',['allclose',['../group__ops.html#gaf0cd4257de7542daf9faf5e605e31020',1,'mlx::core']]], ['allgather_41',['AllGather',['../classmlx_1_1core_1_1distributed_1_1_all_gather.html',1,'mlx::core::distributed::AllGather'],['../classmlx_1_1core_1_1distributed_1_1_all_gather.html#af4b10a5b61f160fb64353057c185b661',1,'mlx::core::distributed::AllGather::AllGather()']]], diff --git a/docs/build/html/search/all_11.js b/docs/build/html/search/all_11.js index 2c8497803..1d12b31b2 100644 --- a/docs/build/html/search/all_11.js +++ b/docs/build/html/search/all_11.js @@ -27,5 +27,6 @@ var searchData= ['queue_24',['queue',['../structmlx_1_1core_1_1metal_1_1_device_stream.html#a77c75a63c51ea56815a86bd882ed190d',1,'mlx::core::metal::DeviceStream']]], ['quiet_5fnan_25',['quiet_NaN',['../structmetal_1_1__numeric__limits__impl_3_01bfloat16__t_01_4.html#aebeb07c01984be246bc2d1b8f8e4ac7b',1,'metal::_numeric_limits_impl< bfloat16_t >']]], ['qvm_26',['qvm',['../quantized_8h.html#ad84f7d5ab9e32dbbe3ca759ae5d5d5c5',1,'quantized.h']]], - ['qvm_5fimpl_27',['qvm_impl',['../quantized_8h.html#a4a8c8db7d5d480733726fd6d1a645e12',1,'quantized.h']]] + ['qvm_5fimpl_27',['qvm_impl',['../quantized_8h.html#a1546533c5b925b2fbb3bec870ec7487a',1,'quantized.h']]], + ['qvm_5fsplit_5fk_28',['qvm_split_k',['../quantized_8h.html#ab8243818512d6078d23e6ffb65fd7bb8',1,'quantized.h']]] ]; diff --git a/docs/build/html/search/all_13.js b/docs/build/html/search/all_13.js index c2d4bb454..38e2a2c7f 100644 --- a/docs/build/html/search/all_13.js +++ b/docs/build/html/search/all_13.js @@ -28,7 +28,7 @@ var searchData= ['scheduler_25',['Scheduler',['../classmlx_1_1core_1_1scheduler_1_1_scheduler.html',1,'mlx::core::scheduler::Scheduler'],['../classmlx_1_1core_1_1scheduler_1_1_scheduler.html#a3ae42aed78a2200e9d02776fcd2316ba',1,'mlx::core::scheduler::Scheduler::Scheduler()'],['../classmlx_1_1core_1_1scheduler_1_1_scheduler.html#a61a74e3628899e66dde600e24a750648',1,'mlx::core::scheduler::Scheduler::Scheduler(const Scheduler &)=delete'],['../classmlx_1_1core_1_1scheduler_1_1_scheduler.html#ac3f77b7c93220dadd0b3bb2e903b7059',1,'mlx::core::scheduler::Scheduler::Scheduler(Scheduler &&)=delete']]], ['scheduler_26',['scheduler',['../namespacemlx_1_1core_1_1scheduler.html#ae856e468c2f7c8f8ec672522cc13730b',1,'mlx::core::scheduler']]], ['scheduler_2eh_27',['scheduler.h',['../scheduler_8h.html',1,'']]], - ['sdpa_5fvector_28',['sdpa_vector',['../sdpa__vector_8h.html#a6f0d7918430064bab910bdaa6c64e927',1,'sdpa_vector.h']]], + ['sdpa_5fvector_28',['sdpa_vector',['../sdpa__vector_8h.html#a4bf36f16e16c1c62d9b243573568e5ae',1,'sdpa_vector.h']]], ['sdpa_5fvector_2eh_29',['sdpa_vector.h',['../sdpa__vector_8h.html',1,'']]], ['seed_30',['seed',['../classmlx_1_1core_1_1random_1_1_key_sequence.html#a9f19c5da2031cba50d0ff996924347d8',1,'mlx::core::random::KeySequence::seed()'],['../namespacemlx_1_1core_1_1random.html#ac4ad325b613257306df74595d3d0e23b',1,'mlx::core::random::seed()']]], ['seek_31',['seek',['../structmlx_1_1core_1_1_contiguous_iterator.html#a24719ee9e8667885d29c2ad74445520c',1,'mlx::core::ContiguousIterator::seek()'],['../classmlx_1_1core_1_1io_1_1_reader.html#acea55078bd39ccaa27a9a36f17a39cd1',1,'mlx::core::io::Reader::seek()'],['../classmlx_1_1core_1_1io_1_1_writer.html#a9c1716dda53aa36faea9c8fb1a3e34d4',1,'mlx::core::io::Writer::seek()'],['../classmlx_1_1core_1_1io_1_1_parallel_file_reader.html#a673c16b669f3cee13f387b7b0a1f39f7',1,'mlx::core::io::ParallelFileReader::seek()'],['../classmlx_1_1core_1_1io_1_1_file_writer.html#a9646f4ea048ae58719daeb588e2de433',1,'mlx::core::io::FileWriter::seek()']]], diff --git a/docs/build/html/search/all_3.js b/docs/build/html/search/all_3.js index 97dcc30db..adac75852 100644 --- a/docs/build/html/search/all_3.js +++ b/docs/build/html/search/all_3.js @@ -35,118 +35,120 @@ var searchData= ['cmplx_3c_20thigh_20_3e_32',['cmplx< Thigh >',['../structpocketfft_1_1detail_1_1cmplx.html',1,'pocketfft::detail']]], ['cndarr_33',['cndarr',['../classpocketfft_1_1detail_1_1cndarr.html',1,'pocketfft::detail::cndarr< T >'],['../classpocketfft_1_1detail_1_1cndarr.html#abf73f1b4ddcfb27d7f85cfa441607129',1,'pocketfft::detail::cndarr::cndarr()']]], ['col_5fcontiguous_34',['col_contiguous',['../structmlx_1_1core_1_1array_1_1_flags.html#ae24709026598d635e6b5c24a15f8a802',1,'mlx::core::array::Flags']]], - ['col_5freduce_5flooped_35',['col_reduce_looped',['../reduce__col_8h.html#a11bfc6112ae2386ac03f5ea7b7d93385',1,'reduce_col.h']]], - ['col_5freduce_5fsmall_36',['col_reduce_small',['../reduce__col_8h.html#adf7aeb18cd1d5042cf6d9b46b582d8ce',1,'reduce_col.h']]], - ['collapse_5fcontiguous_5fdims_37',['collapse_contiguous_dims',['../namespacemlx_1_1core.html#a38fe6ec5220d13d96c7dad7556d2b613',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< std::vector< int64_t > > &strides, int64_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#af2895f9b0083efd8221275eb8cadccbe',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< std::vector< size_t > > &strides, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a90e2b6edc0fe82230cb93f5ea39febb4',1,'mlx::core::collapse_contiguous_dims(const std::vector< array > &xs, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#ac813412cce77fc1340dcfefc6e099276',1,'mlx::core::collapse_contiguous_dims(Arrays &&... xs)'],['../namespacemlx_1_1core.html#aab3cc7f3808934ae0727b920eba231bd',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< int64_t > &strides, int64_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a1e0cbcf109d32794ffc8efc7302ba9b0',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< size_t > &strides, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a4ee50bfb240512d0c0ce151dfe2c74ef',1,'mlx::core::collapse_contiguous_dims(const array &a, size_t size_cap=std::numeric_limits< int32_t >::max())']]], - ['commandencoder_38',['CommandEncoder',['../structmlx_1_1core_1_1metal_1_1_command_encoder.html',1,'mlx::core::metal::CommandEncoder'],['../structmlx_1_1core_1_1metal_1_1_command_encoder.html#a2334774486f447213ee997e55c2e52a3',1,'mlx::core::metal::CommandEncoder::CommandEncoder(MTL::CommandBuffer *cbuf)'],['../structmlx_1_1core_1_1metal_1_1_command_encoder.html#ac68ca977b5bde5434284ce7979647f14',1,'mlx::core::metal::CommandEncoder::CommandEncoder(const CommandEncoder &)=delete']]], - ['commit_5fcommand_5fbuffer_39',['commit_command_buffer',['../classmlx_1_1core_1_1metal_1_1_device.html#a95248f1387824067fd4fed23ace5ac0c',1,'mlx::core::metal::Device']]], - ['commonallocator_40',['CommonAllocator',['../classmlx_1_1core_1_1allocator_1_1_common_allocator.html',1,'mlx::core::allocator']]], - ['communication_5fstream_41',['communication_stream',['../namespacemlx_1_1core_1_1distributed_1_1detail.html#ac3612edf0e0e18c1e4ba0ce7c6e35cd6',1,'mlx::core::distributed::detail']]], - ['compile_42',['compile',['../namespacemlx_1_1core.html#a3ac798e65e59fe10b7fb5c522efce782',1,'mlx::core::compile()'],['../namespacemlx_1_1core_1_1detail.html#ac3b7b09892ff7290d5f3ef26cb444329',1,'mlx::core::detail::compile()']]], - ['compile_2eh_43',['compile.h',['../compile_8h.html',1,'']]], - ['compile_5favailable_5ffor_5fdevice_44',['compile_available_for_device',['../namespacemlx_1_1core_1_1detail.html#aeeff2ba6ec3d9d4ed090de6d2681dbc2',1,'mlx::core::detail']]], - ['compile_5fclear_5fcache_45',['compile_clear_cache',['../namespacemlx_1_1core_1_1detail.html#a3fb927c209b946aefebb195993fbe4cf',1,'mlx::core::detail']]], - ['compile_5ferase_46',['compile_erase',['../namespacemlx_1_1core_1_1detail.html#a69eb76a14f845ca000f1ccb2edda0175',1,'mlx::core::detail']]], - ['compile_5fimpl_2eh_47',['compile_impl.h',['../compile__impl_8h.html',1,'']]], - ['compiled_48',['Compiled',['../classmlx_1_1core_1_1_compiled.html',1,'mlx::core::Compiled'],['../classmlx_1_1core_1_1_compiled.html#a2d8cefff835c419a48a077d306b8e051',1,'mlx::core::Compiled::Compiled()']]], - ['compiled_2eh_49',['compiled.h',['../compiled_8h.html',1,'']]], - ['compiled_5fallocate_5foutputs_50',['compiled_allocate_outputs',['../namespacemlx_1_1core.html#ab8c3c4fc05745f586de922c8266f4fce',1,'mlx::core']]], - ['compiled_5fcheck_5fcontiguity_51',['compiled_check_contiguity',['../namespacemlx_1_1core.html#a3b900ab319948c5a01a3ecd30a709027',1,'mlx::core']]], - ['compiled_5fpreamble_2eh_52',['compiled_preamble.h',['../compiled__preamble_8h.html',1,'']]], - ['compilemode_53',['CompileMode',['../namespacemlx_1_1core.html#adb15ff2b1ca5207fd4f6e631e2c3bcb4',1,'mlx::core']]], - ['complex_2eh_54',['complex.h',['../backend_2metal_2kernels_2complex_8h.html',1,'(Global Namespace)'],['../types_2complex_8h.html',1,'(Global Namespace)']]], - ['complex128_5ft_55',['complex128_t',['../structmlx_1_1core_1_1complex128__t.html',1,'mlx::core::complex128_t'],['../structmlx_1_1core_1_1complex128__t.html#aa15d0b805f8790f7c7b76fc7b9d677e0',1,'mlx::core::complex128_t::complex128_t(double v, double u)'],['../structmlx_1_1core_1_1complex128__t.html#abf2842253b874f9f13f39ea68a89e5b6',1,'mlx::core::complex128_t::complex128_t(std::complex< double > v)'],['../structmlx_1_1core_1_1complex128__t.html#a526fba96d7e815360cb4226af085a1bf',1,'mlx::core::complex128_t::complex128_t(T x)']]], - ['complex64_56',['complex64',['../structmlx_1_1core_1_1_dtype.html#ade845ef5dcebead13a37fe696436e1daa8c022579455bcd2c681f007e84f4e2cf',1,'mlx::core::Dtype::complex64'],['../namespacemlx_1_1core.html#af99db87e0078bfcdb383f5689bc874d4',1,'mlx::core::complex64']]], - ['complex64_5ft_57',['complex64_t',['../structcomplex64__t.html',1,'complex64_t'],['../structmlx_1_1core_1_1complex64__t.html',1,'mlx::core::complex64_t'],['../structcomplex64__t.html#adbd392a5e92d31997380ad0a38be4be8',1,'complex64_t::complex64_t(float real, float imag)'],['../structcomplex64__t.html#a29782289bb90d6294099667b86509cd3',1,'complex64_t::complex64_t()'],['../structcomplex64__t.html#a905b048d70eb8d748a62454268242291',1,'complex64_t::complex64_t() threadgroup'],['../structcomplex64__t.html#a33a2452eb33b5ed53655773539c357a5',1,'complex64_t::complex64_t(T x) thread'],['../structcomplex64__t.html#a89b65ace8588b7bf215355f705eb23d9',1,'complex64_t::complex64_t(T x) threadgroup'],['../structcomplex64__t.html#ac81b486f642fb3b26c5d659917bdbcd0',1,'complex64_t::complex64_t(T x) device'],['../structcomplex64__t.html#a0a27a41206400f1e62b60ceb56960c93',1,'complex64_t::complex64_t(T x) const ant'],['../structmlx_1_1core_1_1complex64__t.html#a697cc973ae27d63c8e00d830e780bd8c',1,'mlx::core::complex64_t::complex64_t(float v, float u)'],['../structmlx_1_1core_1_1complex64__t.html#ae065e39938f9c4374b4116f4c67d4d09',1,'mlx::core::complex64_t::complex64_t(std::complex< float > v)'],['../structmlx_1_1core_1_1complex64__t.html#a2232cbbe591a9d2bc228cb23fac38b50',1,'mlx::core::complex64_t::complex64_t(T x)']]], - ['complex_5fbinop_58',['complex_binop',['../types_2complex_8h.html#a9c7995d495359894e1b30c0f1678d6bd',1,'complex.h']]], - ['complex_5fbinop_5fhelper_59',['complex_binop_helper',['../types_2complex_8h.html#ac6890f9852de12339b09b65757ebc8c4',1,'complex.h']]], - ['complex_5fmul_60',['complex_mul',['../radix_8h.html#a5bfc53b531214c9ce277bebc18aa67d6',1,'radix.h']]], - ['complex_5fmul_5fconj_61',['complex_mul_conj',['../radix_8h.html#a0e2dfd3d1dda09f47ccc64eec35629f3',1,'radix.h']]], - ['complexfloating_62',['complexfloating',['../structmlx_1_1core_1_1_dtype.html#ac091c39cbd6686ef69aa1e5a2425aa2dafb203630099d501ff7c255a574bc4812',1,'mlx::core::Dtype::complexfloating'],['../namespacemlx_1_1core.html#a70b8e88c9df750af984757105af33423',1,'mlx::core::complexfloating']]], - ['compute_5fstrided_5findices_63',['compute_strided_indices',['../struct_read_writer.html#a7c903fbb8b85a856ba5564d7df537cdf',1,'ReadWriter']]], - ['concatenate_64',['Concatenate',['../classmlx_1_1core_1_1_concatenate.html',1,'mlx::core::Concatenate'],['../classmlx_1_1core_1_1_concatenate.html#acff07853de2d31faeec7c4ca40ce0888',1,'mlx::core::Concatenate::Concatenate()']]], - ['concatenate_65',['concatenate',['../group__ops.html#gabdc36fa65697d0361c8d67495de77129',1,'mlx::core::concatenate(const std::vector< array > &arrays, int axis, StreamOrDevice s={})'],['../group__ops.html#gaa95c34ca3a8877f2c50cb60e7fa312b8',1,'mlx::core::concatenate(const std::vector< array > &arrays, StreamOrDevice s={})']]], - ['concatenate_5fgpu_66',['concatenate_gpu',['../namespacemlx_1_1core.html#a050299d0d366ca5c9d09d1004dcc3e7d',1,'mlx::core']]], - ['concurrent_5fqueue_67',['concurrent_queue',['../classpocketfft_1_1detail_1_1threading_1_1concurrent__queue.html',1,'pocketfft::detail::threading']]], - ['concurrent_5fqueue_3c_20std_3a_3afunction_3c_20void_28_29_3e_20_3e_68',['concurrent_queue< std::function< void()> >',['../classpocketfft_1_1detail_1_1threading_1_1concurrent__queue.html',1,'pocketfft::detail::threading']]], - ['concurrentcontext_69',['ConcurrentContext',['../structmlx_1_1core_1_1metal_1_1_command_encoder_1_1_concurrent_context.html',1,'mlx::core::metal::CommandEncoder::ConcurrentContext'],['../structmlx_1_1core_1_1metal_1_1_command_encoder_1_1_concurrent_context.html#aee044d7729739c96e845823f9ecc5174',1,'mlx::core::metal::CommandEncoder::ConcurrentContext::ConcurrentContext()']]], - ['cond_70',['cond',['../structmlx_1_1core_1_1scheduler_1_1_stream_thread.html#a4ffd524d6a5bedd1a303b63bdde6701c',1,'mlx::core::scheduler::StreamThread']]], - ['conj_71',['conj',['../namespacepocketfft_1_1detail.html#a66d79051d502046a9b9f103e744dbad3',1,'pocketfft::detail']]], - ['conjugate_72',['Conjugate',['../struct_conjugate.html',1,'Conjugate'],['../classmlx_1_1core_1_1_conjugate.html',1,'mlx::core::Conjugate'],['../structmlx_1_1core_1_1detail_1_1_conjugate.html',1,'mlx::core::detail::Conjugate'],['../classmlx_1_1core_1_1_conjugate.html#a627f9e6a8729fb3ffb3ca3228d007c87',1,'mlx::core::Conjugate::Conjugate()']]], - ['conjugate_73',['conjugate',['../group__ops.html#ga5b596906bf8cdc8d97ed6ddc9aeb4c23',1,'mlx::core']]], - ['contiguous_74',['contiguous',['../structmlx_1_1core_1_1array_1_1_flags.html#afd0ab11e7a486a2a8e50ee84b971ac8a',1,'mlx::core::array::Flags']]], - ['contiguous_5fscan_75',['contiguous_scan',['../scan_8h.html#a60d279b9add7d56639bb209408f09d79',1,'scan.h']]], - ['contiguousallreduce_76',['ContiguousAllReduce',['../namespacemlx_1_1core.html#a12412984a1cabfe1189942c898f8fe65ae4e34c7154eb8dc47aa8503209730424',1,'mlx::core']]], - ['contiguousiterator_77',['ContiguousIterator',['../structmlx_1_1core_1_1_contiguous_iterator.html',1,'mlx::core::ContiguousIterator< StrideT >'],['../structmlx_1_1core_1_1_contiguous_iterator.html#a68794af4a442d3d8ac4647817af8e1f6',1,'mlx::core::ContiguousIterator::ContiguousIterator()'],['../structmlx_1_1core_1_1_contiguous_iterator.html#a6cb378408b6f546eeb6ade1a4faafe3c',1,'mlx::core::ContiguousIterator::ContiguousIterator(const array &a)'],['../structmlx_1_1core_1_1_contiguous_iterator.html#a16bdacb53f65b7284068cd49d4cba292',1,'mlx::core::ContiguousIterator::ContiguousIterator(const std::vector< int > &shape, const std::vector< StrideT > &strides, int dims)']]], - ['contiguousreduce_78',['ContiguousReduce',['../namespacemlx_1_1core.html#a12412984a1cabfe1189942c898f8fe65ad2547f25dffe8d8936dbec25601cfc84',1,'mlx::core']]], - ['contiguousstridedreduce_79',['ContiguousStridedReduce',['../namespacemlx_1_1core.html#a12412984a1cabfe1189942c898f8fe65ab48dac7508a2c790de1bdc33f29177ed',1,'mlx::core']]], - ['conv_80',['conv',['../namespacemlx_1_1core_1_1metal.html#ab1704e853394c725668c06752ebb5c24',1,'mlx::core::metal']]], - ['conv_2eh_81',['conv.h',['../conv_8h.html',1,'']]], - ['conv1d_82',['conv1d',['../group__ops.html#ga30d47e08093c03a3676f235f9f559411',1,'mlx::core']]], - ['conv2d_83',['conv2d',['../group__ops.html#ga73b02833229678786e7f302d458d5a83',1,'mlx::core']]], - ['conv2dgeneralbaseinfo_84',['Conv2DGeneralBaseInfo',['../structmlx_1_1steel_1_1_conv2_d_general_base_info.html',1,'mlx::steel']]], - ['conv2dgeneraljumpparams_85',['Conv2DGeneralJumpParams',['../structmlx_1_1steel_1_1_conv2_d_general_jump_params.html',1,'mlx::steel']]], - ['conv2dinputblockloadergeneral_86',['Conv2DInputBlockLoaderGeneral',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_general.html',1,'mlx::steel::Conv2DInputBlockLoaderGeneral< T, BM, BN, BK, tgp_size, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_general.html#a1d83af561a483432bf8dcb42e734b23b',1,'mlx::steel::Conv2DInputBlockLoaderGeneral::Conv2DInputBlockLoaderGeneral()']]], - ['conv2dinputblockloaderlargefilter_87',['Conv2DInputBlockLoaderLargeFilter',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_large_filter.html',1,'mlx::steel::Conv2DInputBlockLoaderLargeFilter< T, BM, BN, BK, tgp_size, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_large_filter.html#a8755116a535539744e4947bc69f9c50f',1,'mlx::steel::Conv2DInputBlockLoaderLargeFilter::Conv2DInputBlockLoaderLargeFilter()']]], - ['conv2dinputblockloadersmallchannels_88',['Conv2DInputBlockLoaderSmallChannels',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_channels.html',1,'mlx::steel::Conv2DInputBlockLoaderSmallChannels< T, BM, BN, BK, tgp_size, n_channels, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_channels.html#ab9fd3fdeab94470dde3326f1dd5c455a',1,'mlx::steel::Conv2DInputBlockLoaderSmallChannels::Conv2DInputBlockLoaderSmallChannels()']]], - ['conv2dinputblockloadersmallfilter_89',['Conv2DInputBlockLoaderSmallFilter',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_filter.html',1,'mlx::steel::Conv2DInputBlockLoaderSmallFilter< T, BM, BN, BK, tgp_size, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_filter.html#a0a2cbf57c51cd928722e3f06aafcf933',1,'mlx::steel::Conv2DInputBlockLoaderSmallFilter::Conv2DInputBlockLoaderSmallFilter()']]], - ['conv2dweightblockloader_90',['Conv2DWeightBlockLoader',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader.html',1,'mlx::steel::Conv2DWeightBlockLoader< T, BM, BN, BK, tgp_size, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader.html#a9a7dca3512b64cffb6eac305d795831c',1,'mlx::steel::Conv2DWeightBlockLoader::Conv2DWeightBlockLoader()']]], - ['conv2dweightblockloadergeneral_91',['Conv2DWeightBlockLoaderGeneral',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_general.html',1,'mlx::steel::Conv2DWeightBlockLoaderGeneral< T, BM, BN, BK, tgp_size, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_general.html#ad0550fabbdc9297559381a5b488e9af1',1,'mlx::steel::Conv2DWeightBlockLoaderGeneral::Conv2DWeightBlockLoaderGeneral()']]], - ['conv2dweightblockloadersmallchannels_92',['Conv2DWeightBlockLoaderSmallChannels',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_small_channels.html',1,'mlx::steel::Conv2DWeightBlockLoaderSmallChannels< T, BM, BN, BK, tgp_size, n_channels, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_small_channels.html#ae1806ea1c19713819dee83a38ab35fa6',1,'mlx::steel::Conv2DWeightBlockLoaderSmallChannels::Conv2DWeightBlockLoaderSmallChannels()']]], - ['conv3d_93',['conv3d',['../group__ops.html#ga6e9907d2f14dc4803e4306b3dbc4b3ca',1,'mlx::core']]], - ['conv_5fgeneral_94',['conv_general',['../group__ops.html#ga2236e5dfc7e52e28abf6c21675d0a51e',1,'mlx::core::conv_general(array input, array weight, std::vector< int > stride={}, std::vector< int > padding_lo={}, std::vector< int > padding_hi={}, std::vector< int > kernel_dilation={}, std::vector< int > input_dilation={}, int groups=1, bool flip=false, StreamOrDevice s={})'],['../group__ops.html#gab59f89942cd1efaadffe9e8762e3c99d',1,'mlx::core::conv_general(const array &input, const array &weight, std::vector< int > stride={}, std::vector< int > padding={}, std::vector< int > kernel_dilation={}, std::vector< int > input_dilation={}, int groups=1, bool flip=false, StreamOrDevice s={})']]], - ['conv_5ftranspose1d_95',['conv_transpose1d',['../group__ops.html#gaa30bf1adcd78d1c2595d07b215731714',1,'mlx::core']]], - ['conv_5ftranspose2d_96',['conv_transpose2d',['../group__ops.html#gaebb59971cb9bc45005dc1d398e4f0a3d',1,'mlx::core']]], - ['conv_5ftranspose3d_97',['conv_transpose3d',['../group__ops.html#ga8db814da631d9cd32a8d6563bf4ac530',1,'mlx::core']]], - ['convolution_98',['Convolution',['../classmlx_1_1core_1_1_convolution.html',1,'mlx::core::Convolution'],['../classmlx_1_1core_1_1_convolution.html#a6f1de77b719bb13217b0d8c64cabb8ef',1,'mlx::core::Convolution::Convolution()']]], - ['copy_99',['Copy',['../classmlx_1_1core_1_1_copy.html',1,'mlx::core::Copy'],['../classmlx_1_1core_1_1_copy.html#a6243e044af119105ffaaed7d405cd584',1,'mlx::core::Copy::Copy()']]], - ['copy_100',['copy',['../namespacemlx_1_1core.html#a479648542a2bea151b947b18f0e79dd2',1,'mlx::core::copy()'],['../namespacemlx_1_1core_1_1metal.html#aa215e631e2680f04a591b88d91571719',1,'mlx::core::metal::copy()'],['../group__ops.html#gae306e93af12f774bd80bad6c231b09d6',1,'mlx::core::copy()']]], - ['copy_2eh_101',['copy.h',['../common_2copy_8h.html',1,'(Global Namespace)'],['../metal_2copy_8h.html',1,'(Global Namespace)'],['../metal_2kernels_2copy_8h.html',1,'(Global Namespace)']]], - ['copy_5fg_102',['copy_g',['../metal_2kernels_2copy_8h.html#a778ce2dbfbaa23b24bd5efbe68448c36',1,'copy.h']]], - ['copy_5fg_5fnd1_103',['copy_g_nd1',['../metal_2kernels_2copy_8h.html#aba4530a7db6a61ca36f50e4f5e58fb77',1,'copy.h']]], - ['copy_5fg_5fnd2_104',['copy_g_nd2',['../metal_2kernels_2copy_8h.html#aee678c7c31119f3e609685589f37490c',1,'copy.h']]], - ['copy_5fg_5fnd3_105',['copy_g_nd3',['../metal_2kernels_2copy_8h.html#a821f8f3f3891159a295c66fc25aed1ff',1,'copy.h']]], - ['copy_5fgg_106',['copy_gg',['../metal_2kernels_2copy_8h.html#a1e39c2683eeaf05955e7619fbd34aea5',1,'copy.h']]], - ['copy_5fgg_5fnd1_107',['copy_gg_nd1',['../metal_2kernels_2copy_8h.html#a3278d9c999718bee3ccbe2922f501bf1',1,'copy.h']]], - ['copy_5fgg_5fnd2_108',['copy_gg_nd2',['../metal_2kernels_2copy_8h.html#a3e2d3cc7f34f56170409b6735f51a950',1,'copy.h']]], - ['copy_5fgg_5fnd3_109',['copy_gg_nd3',['../metal_2kernels_2copy_8h.html#a59f43b5bffed936d7559ceb06a10aabd',1,'copy.h']]], - ['copy_5fgpu_110',['copy_gpu',['../namespacemlx_1_1core.html#addaa46a13ac2deb1d9ce621338320e0e',1,'mlx::core::copy_gpu(const array &src, array &out, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#a6a6f4e46c8fc44fdc74c50ace02bcf38',1,'mlx::core::copy_gpu(const array &src, array &out, CopyType ctype)']]], - ['copy_5fgpu_5finplace_111',['copy_gpu_inplace',['../namespacemlx_1_1core.html#a69e30f5d30a6d72ac0ffe4886f24b7ba',1,'mlx::core::copy_gpu_inplace(const array &in, array &out, const std::vector< int > &data_shape, const std::vector< stride_t > &i_strides, const std::vector< stride_t > &o_strides, int64_t i_offset, int64_t o_offset, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#a8e1ccb0ed9387b0a789311d9f8964803',1,'mlx::core::copy_gpu_inplace(const array &src, array &out, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#ae55b801b09ccf55cba96278163a9b1ef',1,'mlx::core::copy_gpu_inplace(const array &in, array &out, const std::vector< int64_t > &istride, int64_t ioffset, CopyType ctype, const Stream &s)']]], - ['copy_5fhartley_112',['copy_hartley',['../namespacepocketfft_1_1detail.html#abac3fcc8ce83800d228774f64c28d4c3',1,'pocketfft::detail::copy_hartley(const multi_iter< vlen > &it, const vtype_t< T > *src, ndarr< T > &dst)'],['../namespacepocketfft_1_1detail.html#ae7b44d2773d9d06a9787aff01d66b3ed',1,'pocketfft::detail::copy_hartley(const multi_iter< vlen > &it, const T *src, ndarr< T > &dst)']]], - ['copy_5finplace_113',['copy_inplace',['../namespacemlx_1_1core.html#a98495894a796b2cc6d022e7a03432c64',1,'mlx::core::copy_inplace(const array &src, array &dst, CopyType ctype)'],['../namespacemlx_1_1core.html#aad636e2d0b2f882cadd1b438f4daa9ed',1,'mlx::core::copy_inplace(const array &src, array &dst, const std::vector< int > &data_shape, const std::vector< stride_t > &i_strides, const std::vector< stride_t > &o_strides, int64_t i_offset, int64_t o_offset, CopyType ctype)']]], - ['copy_5finput_114',['copy_input',['../namespacepocketfft_1_1detail.html#aff05be3064743c1143b19318ab12ad4a',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< cmplx< T > > &src, cmplx< vtype_t< T > > *dst)'],['../namespacepocketfft_1_1detail.html#a30fc708f9d8f9cfa74194925c7863c0a',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< T > &src, vtype_t< T > *dst)'],['../namespacepocketfft_1_1detail.html#a3387bd35f237870e42b8461769e6aec4',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< T > &src, T *dst)']]], - ['copy_5foutput_115',['copy_output',['../namespacepocketfft_1_1detail.html#a1523a037300a8da05db210b802d9cb0e',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const cmplx< vtype_t< T > > *src, ndarr< cmplx< T > > &dst)'],['../namespacepocketfft_1_1detail.html#a21980853aca4d92ed06e3dcffe7ef660',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const vtype_t< T > *src, ndarr< T > &dst)'],['../namespacepocketfft_1_1detail.html#a310481c334e46674710ba794ad7403c0',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const T *src, ndarr< T > &dst)']]], - ['copy_5fs_116',['copy_s',['../metal_2kernels_2copy_8h.html#aef09f9b9475345b1bba121d037d222ea',1,'copy.h']]], - ['copy_5fs2_117',['copy_s2',['../metal_2kernels_2copy_8h.html#a8023e9335cc5334847a8d315042be3a3',1,'copy.h']]], - ['copy_5fshared_5fbuffer_118',['copy_shared_buffer',['../classmlx_1_1core_1_1array.html#a28df7a333d90a311c49bc4bce7a1ad6d',1,'mlx::core::array::copy_shared_buffer(const array &other, const std::vector< size_t > &strides, Flags flags, size_t data_size, size_t offset=0)'],['../classmlx_1_1core_1_1array.html#a92974c656c35a972ad241f80584bbd29',1,'mlx::core::array::copy_shared_buffer(const array &other)']]], - ['copy_5fv_119',['copy_v',['../metal_2kernels_2copy_8h.html#ae26a13e0c8e6c15f7b10078e65970659',1,'copy.h']]], - ['copy_5fv2_120',['copy_v2',['../metal_2kernels_2copy_8h.html#aee14a5326f53d9b30b0b38e27d180ef3',1,'copy.h']]], - ['copytype_121',['CopyType',['../namespacemlx_1_1core.html#abd84ff6c5245e4e170b2ef5247594337',1,'mlx::core']]], - ['core_20array_20operations_122',['Core array operations',['../group__ops.html',1,'']]], - ['cos_123',['Cos',['../struct_cos.html',1,'Cos'],['../classmlx_1_1core_1_1_cos.html',1,'mlx::core::Cos'],['../structmlx_1_1core_1_1detail_1_1_cos.html',1,'mlx::core::detail::Cos'],['../classmlx_1_1core_1_1_cos.html#a2acb9fcf0901462189c476756fd99995',1,'mlx::core::Cos::Cos()']]], - ['cos_124',['cos',['../namespacepocketfft_1_1detail.html#a499c1e8b7d79a5272af024f46c63ff9d',1,'pocketfft::detail::cos()'],['../namespacemetal.html#a2fa4778a6fe2fa43253ea724e5a608a3',1,'metal::cos()'],['../namespacemetal_1_1fast.html#a75b6bb32fa3870eda46a7bfc9f481f88',1,'metal::fast::cos()'],['../namespacemetal_1_1precise.html#ac4941f62e7d8ab9d7cabbd967aa9f220',1,'metal::precise::cos()'],['../group__ops.html#ga39dfdf72b556012aa35ff27a94116e74',1,'mlx::core::cos()']]], - ['cosh_125',['Cosh',['../struct_cosh.html',1,'Cosh'],['../classmlx_1_1core_1_1_cosh.html',1,'mlx::core::Cosh'],['../structmlx_1_1core_1_1detail_1_1_cosh.html',1,'mlx::core::detail::Cosh'],['../classmlx_1_1core_1_1_cosh.html#a44e8ac2e09a55ec32e9dc6641eedc8f1',1,'mlx::core::Cosh::Cosh()']]], - ['cosh_126',['cosh',['../namespacemetal.html#a8a68a88cc110830d057dbd71431b93c0',1,'metal::cosh()'],['../namespacemetal_1_1fast.html#a31544ad9de28012a4ddda86e3966a77e',1,'metal::fast::cosh()'],['../namespacemetal_1_1precise.html#a72d86d508300a9b58f4ccbbe70da4fbc',1,'metal::precise::cosh()'],['../group__ops.html#ga2181b71cda88007a3092be4795ff0715',1,'mlx::core::cosh()']]], - ['cosine_127',['cosine',['../structpocketfft_1_1detail_1_1_exec_dcst.html#a185023fc1e386cc8f233b79c49c1fd8a',1,'pocketfft::detail::ExecDcst']]], - ['cospi_128',['cospi',['../namespacemetal.html#a5c2f37939ad705ddea4409d3bedb8ce1',1,'metal::cospi()'],['../namespacemetal_1_1fast.html#a9906b41f75319b384ffb570cc94d67ce',1,'metal::fast::cospi()'],['../namespacemetal_1_1precise.html#a2392b78bd196efdbbac65901c4ab20e7',1,'metal::precise::cospi()']]], - ['cost_5fguess_129',['cost_guess',['../structpocketfft_1_1detail_1_1util.html#ad3d874bc3fb0048df2270779a15d4bd0',1,'pocketfft::detail::util']]], - ['count_5fdown_130',['count_down',['../classpocketfft_1_1detail_1_1threading_1_1latch.html#a81d6597189b40410e35f3cd653fd1342',1,'pocketfft::detail::threading::latch']]], - ['cpu_131',['cpu',['../structmlx_1_1core_1_1_device.html#a69ee81924251dec96f1945c9d91506fd',1,'mlx::core::Device::cpu'],['../structmlx_1_1core_1_1_device.html#ac45b3de9b3458d8f31005136cde20fdbad9747e2da342bdb995f6389533ad1a3d',1,'mlx::core::Device::cpu']]], - ['cross_132',['cross',['../namespacemlx_1_1core_1_1linalg.html#abcda3fbda45183c21e7f27aa0dde64e6',1,'mlx::core::linalg']]], - ['ctile_133',['Ctile',['../structmlx_1_1steel_1_1_block_m_m_a.html#a81838da5d81e62d372d581be599c5a88',1,'mlx::steel::BlockMMA']]], - ['cummax_134',['CumMax',['../struct_cum_max.html',1,'']]], - ['cummax_135',['cummax',['../group__ops.html#gaee37cac8476e8f8d666bcded5bc59143',1,'mlx::core']]], - ['cummin_136',['CumMin',['../struct_cum_min.html',1,'']]], - ['cummin_137',['cummin',['../group__ops.html#ga19c1bf6929fe8d66b9cd408946aea6a8',1,'mlx::core']]], - ['cumprod_138',['CumProd',['../struct_cum_prod.html',1,'']]], - ['cumprod_139',['cumprod',['../group__ops.html#ga0d71dfbc14ef3ed564b0c5ee26af680f',1,'mlx::core']]], - ['cumprod_3c_20bool_20_3e_140',['CumProd< bool >',['../struct_cum_prod_3_01bool_01_4.html',1,'']]], - ['cumsum_141',['CumSum',['../struct_cum_sum.html',1,'']]], - ['cumsum_142',['cumsum',['../group__ops.html#gaddc825a5c173e195ab0fda83ad630420',1,'mlx::core']]], - ['custom_143',['Custom',['../classmlx_1_1core_1_1fast_1_1_custom.html',1,'mlx::core::fast::Custom'],['../classmlx_1_1core_1_1fast_1_1_custom.html#a4186fea23f7156c38960426821fca313',1,'mlx::core::fast::Custom::Custom()']]], - ['custom_5ffunction_144',['custom_function',['../namespacemlx_1_1core.html#a8d3ca5fbaecdb995660c24cde5aeebaf',1,'mlx::core']]], - ['custom_5fvjp_145',['custom_vjp',['../namespacemlx_1_1core.html#a9290596250fa308df4c69b44483bb8aa',1,'mlx::core']]], - ['customkernel_146',['CustomKernel',['../classmlx_1_1core_1_1fast_1_1_custom_kernel.html',1,'mlx::core::fast::CustomKernel'],['../classmlx_1_1core_1_1fast_1_1_custom_kernel.html#a954893e07f0d36715b4e1e414b6f2153',1,'mlx::core::fast::CustomKernel::CustomKernel()']]], - ['customkernelshapeinfo_147',['CustomKernelShapeInfo',['../structmlx_1_1core_1_1fast_1_1_custom_kernel_shape_info.html',1,'mlx::core::fast']]], - ['customtransforms_148',['CustomTransforms',['../classmlx_1_1core_1_1_custom_transforms.html',1,'mlx::core::CustomTransforms'],['../classmlx_1_1core_1_1_custom_transforms.html#ab52abadb9c6f6db83d087c7b751be488',1,'mlx::core::CustomTransforms::CustomTransforms()']]] + ['col_5freduce_5f2pass_35',['col_reduce_2pass',['../reduce__col_8h.html#a0e92fc74eeaa8ee2ceb83bafc6eb1d7d',1,'reduce_col.h']]], + ['col_5freduce_5flongcolumn_36',['col_reduce_longcolumn',['../reduce__col_8h.html#a5b4f4c4c247ad341ff8d31dcbbbce0eb',1,'reduce_col.h']]], + ['col_5freduce_5flooped_37',['col_reduce_looped',['../reduce__col_8h.html#a11bfc6112ae2386ac03f5ea7b7d93385',1,'reduce_col.h']]], + ['col_5freduce_5fsmall_38',['col_reduce_small',['../reduce__col_8h.html#a7c378443a2b6f4d9210db8a21a9ac4f5',1,'reduce_col.h']]], + ['collapse_5fcontiguous_5fdims_39',['collapse_contiguous_dims',['../namespacemlx_1_1core.html#a38fe6ec5220d13d96c7dad7556d2b613',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< std::vector< int64_t > > &strides, int64_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#af2895f9b0083efd8221275eb8cadccbe',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< std::vector< size_t > > &strides, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a90e2b6edc0fe82230cb93f5ea39febb4',1,'mlx::core::collapse_contiguous_dims(const std::vector< array > &xs, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#ac813412cce77fc1340dcfefc6e099276',1,'mlx::core::collapse_contiguous_dims(Arrays &&... xs)'],['../namespacemlx_1_1core.html#aab3cc7f3808934ae0727b920eba231bd',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< int64_t > &strides, int64_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a1e0cbcf109d32794ffc8efc7302ba9b0',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< size_t > &strides, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a4ee50bfb240512d0c0ce151dfe2c74ef',1,'mlx::core::collapse_contiguous_dims(const array &a, size_t size_cap=std::numeric_limits< int32_t >::max())']]], + ['commandencoder_40',['CommandEncoder',['../structmlx_1_1core_1_1metal_1_1_command_encoder.html',1,'mlx::core::metal::CommandEncoder'],['../structmlx_1_1core_1_1metal_1_1_command_encoder.html#a2334774486f447213ee997e55c2e52a3',1,'mlx::core::metal::CommandEncoder::CommandEncoder(MTL::CommandBuffer *cbuf)'],['../structmlx_1_1core_1_1metal_1_1_command_encoder.html#ac68ca977b5bde5434284ce7979647f14',1,'mlx::core::metal::CommandEncoder::CommandEncoder(const CommandEncoder &)=delete']]], + ['commit_5fcommand_5fbuffer_41',['commit_command_buffer',['../classmlx_1_1core_1_1metal_1_1_device.html#a95248f1387824067fd4fed23ace5ac0c',1,'mlx::core::metal::Device']]], + ['commonallocator_42',['CommonAllocator',['../classmlx_1_1core_1_1allocator_1_1_common_allocator.html',1,'mlx::core::allocator']]], + ['communication_5fstream_43',['communication_stream',['../namespacemlx_1_1core_1_1distributed_1_1detail.html#ac3612edf0e0e18c1e4ba0ce7c6e35cd6',1,'mlx::core::distributed::detail']]], + ['compile_44',['compile',['../namespacemlx_1_1core.html#a3ac798e65e59fe10b7fb5c522efce782',1,'mlx::core::compile()'],['../namespacemlx_1_1core_1_1detail.html#ac3b7b09892ff7290d5f3ef26cb444329',1,'mlx::core::detail::compile()']]], + ['compile_2eh_45',['compile.h',['../compile_8h.html',1,'']]], + ['compile_5favailable_5ffor_5fdevice_46',['compile_available_for_device',['../namespacemlx_1_1core_1_1detail.html#aeeff2ba6ec3d9d4ed090de6d2681dbc2',1,'mlx::core::detail']]], + ['compile_5fclear_5fcache_47',['compile_clear_cache',['../namespacemlx_1_1core_1_1detail.html#a3fb927c209b946aefebb195993fbe4cf',1,'mlx::core::detail']]], + ['compile_5ferase_48',['compile_erase',['../namespacemlx_1_1core_1_1detail.html#a69eb76a14f845ca000f1ccb2edda0175',1,'mlx::core::detail']]], + ['compile_5fimpl_2eh_49',['compile_impl.h',['../compile__impl_8h.html',1,'']]], + ['compiled_50',['Compiled',['../classmlx_1_1core_1_1_compiled.html',1,'mlx::core::Compiled'],['../classmlx_1_1core_1_1_compiled.html#a2d8cefff835c419a48a077d306b8e051',1,'mlx::core::Compiled::Compiled()']]], + ['compiled_2eh_51',['compiled.h',['../compiled_8h.html',1,'']]], + ['compiled_5fallocate_5foutputs_52',['compiled_allocate_outputs',['../namespacemlx_1_1core.html#ab8c3c4fc05745f586de922c8266f4fce',1,'mlx::core']]], + ['compiled_5fcheck_5fcontiguity_53',['compiled_check_contiguity',['../namespacemlx_1_1core.html#a3b900ab319948c5a01a3ecd30a709027',1,'mlx::core']]], + ['compiled_5fpreamble_2eh_54',['compiled_preamble.h',['../compiled__preamble_8h.html',1,'']]], + ['compilemode_55',['CompileMode',['../namespacemlx_1_1core.html#adb15ff2b1ca5207fd4f6e631e2c3bcb4',1,'mlx::core']]], + ['complex_2eh_56',['complex.h',['../backend_2metal_2kernels_2complex_8h.html',1,'(Global Namespace)'],['../types_2complex_8h.html',1,'(Global Namespace)']]], + ['complex128_5ft_57',['complex128_t',['../structmlx_1_1core_1_1complex128__t.html',1,'mlx::core::complex128_t'],['../structmlx_1_1core_1_1complex128__t.html#aa15d0b805f8790f7c7b76fc7b9d677e0',1,'mlx::core::complex128_t::complex128_t(double v, double u)'],['../structmlx_1_1core_1_1complex128__t.html#abf2842253b874f9f13f39ea68a89e5b6',1,'mlx::core::complex128_t::complex128_t(std::complex< double > v)'],['../structmlx_1_1core_1_1complex128__t.html#a526fba96d7e815360cb4226af085a1bf',1,'mlx::core::complex128_t::complex128_t(T x)']]], + ['complex64_58',['complex64',['../structmlx_1_1core_1_1_dtype.html#ade845ef5dcebead13a37fe696436e1daa8c022579455bcd2c681f007e84f4e2cf',1,'mlx::core::Dtype::complex64'],['../namespacemlx_1_1core.html#af99db87e0078bfcdb383f5689bc874d4',1,'mlx::core::complex64']]], + ['complex64_5ft_59',['complex64_t',['../structcomplex64__t.html',1,'complex64_t'],['../structmlx_1_1core_1_1complex64__t.html',1,'mlx::core::complex64_t'],['../structcomplex64__t.html#adbd392a5e92d31997380ad0a38be4be8',1,'complex64_t::complex64_t(float real, float imag)'],['../structcomplex64__t.html#a29782289bb90d6294099667b86509cd3',1,'complex64_t::complex64_t()'],['../structcomplex64__t.html#a905b048d70eb8d748a62454268242291',1,'complex64_t::complex64_t() threadgroup'],['../structcomplex64__t.html#a33a2452eb33b5ed53655773539c357a5',1,'complex64_t::complex64_t(T x) thread'],['../structcomplex64__t.html#a89b65ace8588b7bf215355f705eb23d9',1,'complex64_t::complex64_t(T x) threadgroup'],['../structcomplex64__t.html#ac81b486f642fb3b26c5d659917bdbcd0',1,'complex64_t::complex64_t(T x) device'],['../structcomplex64__t.html#a0a27a41206400f1e62b60ceb56960c93',1,'complex64_t::complex64_t(T x) const ant'],['../structmlx_1_1core_1_1complex64__t.html#a697cc973ae27d63c8e00d830e780bd8c',1,'mlx::core::complex64_t::complex64_t(float v, float u)'],['../structmlx_1_1core_1_1complex64__t.html#ae065e39938f9c4374b4116f4c67d4d09',1,'mlx::core::complex64_t::complex64_t(std::complex< float > v)'],['../structmlx_1_1core_1_1complex64__t.html#a2232cbbe591a9d2bc228cb23fac38b50',1,'mlx::core::complex64_t::complex64_t(T x)']]], + ['complex_5fbinop_60',['complex_binop',['../types_2complex_8h.html#a9c7995d495359894e1b30c0f1678d6bd',1,'complex.h']]], + ['complex_5fbinop_5fhelper_61',['complex_binop_helper',['../types_2complex_8h.html#ac6890f9852de12339b09b65757ebc8c4',1,'complex.h']]], + ['complex_5fmul_62',['complex_mul',['../radix_8h.html#a5bfc53b531214c9ce277bebc18aa67d6',1,'radix.h']]], + ['complex_5fmul_5fconj_63',['complex_mul_conj',['../radix_8h.html#a0e2dfd3d1dda09f47ccc64eec35629f3',1,'radix.h']]], + ['complexfloating_64',['complexfloating',['../structmlx_1_1core_1_1_dtype.html#ac091c39cbd6686ef69aa1e5a2425aa2dafb203630099d501ff7c255a574bc4812',1,'mlx::core::Dtype::complexfloating'],['../namespacemlx_1_1core.html#a70b8e88c9df750af984757105af33423',1,'mlx::core::complexfloating']]], + ['compute_5fstrided_5findices_65',['compute_strided_indices',['../struct_read_writer.html#a7c903fbb8b85a856ba5564d7df537cdf',1,'ReadWriter']]], + ['concatenate_66',['Concatenate',['../classmlx_1_1core_1_1_concatenate.html',1,'mlx::core::Concatenate'],['../classmlx_1_1core_1_1_concatenate.html#acff07853de2d31faeec7c4ca40ce0888',1,'mlx::core::Concatenate::Concatenate()']]], + ['concatenate_67',['concatenate',['../group__ops.html#gabdc36fa65697d0361c8d67495de77129',1,'mlx::core::concatenate(const std::vector< array > &arrays, int axis, StreamOrDevice s={})'],['../group__ops.html#gaa95c34ca3a8877f2c50cb60e7fa312b8',1,'mlx::core::concatenate(const std::vector< array > &arrays, StreamOrDevice s={})']]], + ['concatenate_5fgpu_68',['concatenate_gpu',['../namespacemlx_1_1core.html#a050299d0d366ca5c9d09d1004dcc3e7d',1,'mlx::core']]], + ['concurrent_5fqueue_69',['concurrent_queue',['../classpocketfft_1_1detail_1_1threading_1_1concurrent__queue.html',1,'pocketfft::detail::threading']]], + ['concurrent_5fqueue_3c_20std_3a_3afunction_3c_20void_28_29_3e_20_3e_70',['concurrent_queue< std::function< void()> >',['../classpocketfft_1_1detail_1_1threading_1_1concurrent__queue.html',1,'pocketfft::detail::threading']]], + ['concurrentcontext_71',['ConcurrentContext',['../structmlx_1_1core_1_1metal_1_1_command_encoder_1_1_concurrent_context.html',1,'mlx::core::metal::CommandEncoder::ConcurrentContext'],['../structmlx_1_1core_1_1metal_1_1_command_encoder_1_1_concurrent_context.html#aee044d7729739c96e845823f9ecc5174',1,'mlx::core::metal::CommandEncoder::ConcurrentContext::ConcurrentContext()']]], + ['cond_72',['cond',['../structmlx_1_1core_1_1scheduler_1_1_stream_thread.html#a4ffd524d6a5bedd1a303b63bdde6701c',1,'mlx::core::scheduler::StreamThread']]], + ['conj_73',['conj',['../namespacepocketfft_1_1detail.html#a66d79051d502046a9b9f103e744dbad3',1,'pocketfft::detail']]], + ['conjugate_74',['Conjugate',['../struct_conjugate.html',1,'Conjugate'],['../classmlx_1_1core_1_1_conjugate.html',1,'mlx::core::Conjugate'],['../structmlx_1_1core_1_1detail_1_1_conjugate.html',1,'mlx::core::detail::Conjugate'],['../classmlx_1_1core_1_1_conjugate.html#a627f9e6a8729fb3ffb3ca3228d007c87',1,'mlx::core::Conjugate::Conjugate()']]], + ['conjugate_75',['conjugate',['../group__ops.html#ga5b596906bf8cdc8d97ed6ddc9aeb4c23',1,'mlx::core']]], + ['contiguous_76',['contiguous',['../structmlx_1_1core_1_1array_1_1_flags.html#afd0ab11e7a486a2a8e50ee84b971ac8a',1,'mlx::core::array::Flags']]], + ['contiguous_5fscan_77',['contiguous_scan',['../scan_8h.html#a60d279b9add7d56639bb209408f09d79',1,'scan.h']]], + ['contiguousallreduce_78',['ContiguousAllReduce',['../namespacemlx_1_1core.html#a12412984a1cabfe1189942c898f8fe65ae4e34c7154eb8dc47aa8503209730424',1,'mlx::core']]], + ['contiguousiterator_79',['ContiguousIterator',['../structmlx_1_1core_1_1_contiguous_iterator.html',1,'mlx::core::ContiguousIterator< StrideT >'],['../structmlx_1_1core_1_1_contiguous_iterator.html#a68794af4a442d3d8ac4647817af8e1f6',1,'mlx::core::ContiguousIterator::ContiguousIterator()'],['../structmlx_1_1core_1_1_contiguous_iterator.html#a6cb378408b6f546eeb6ade1a4faafe3c',1,'mlx::core::ContiguousIterator::ContiguousIterator(const array &a)'],['../structmlx_1_1core_1_1_contiguous_iterator.html#a16bdacb53f65b7284068cd49d4cba292',1,'mlx::core::ContiguousIterator::ContiguousIterator(const std::vector< int > &shape, const std::vector< StrideT > &strides, int dims)']]], + ['contiguousreduce_80',['ContiguousReduce',['../namespacemlx_1_1core.html#a12412984a1cabfe1189942c898f8fe65ad2547f25dffe8d8936dbec25601cfc84',1,'mlx::core']]], + ['contiguousstridedreduce_81',['ContiguousStridedReduce',['../namespacemlx_1_1core.html#a12412984a1cabfe1189942c898f8fe65ab48dac7508a2c790de1bdc33f29177ed',1,'mlx::core']]], + ['conv_82',['conv',['../namespacemlx_1_1core_1_1metal.html#ab1704e853394c725668c06752ebb5c24',1,'mlx::core::metal']]], + ['conv_2eh_83',['conv.h',['../conv_8h.html',1,'']]], + ['conv1d_84',['conv1d',['../group__ops.html#ga30d47e08093c03a3676f235f9f559411',1,'mlx::core']]], + ['conv2d_85',['conv2d',['../group__ops.html#ga73b02833229678786e7f302d458d5a83',1,'mlx::core']]], + ['conv2dgeneralbaseinfo_86',['Conv2DGeneralBaseInfo',['../structmlx_1_1steel_1_1_conv2_d_general_base_info.html',1,'mlx::steel']]], + ['conv2dgeneraljumpparams_87',['Conv2DGeneralJumpParams',['../structmlx_1_1steel_1_1_conv2_d_general_jump_params.html',1,'mlx::steel']]], + ['conv2dinputblockloadergeneral_88',['Conv2DInputBlockLoaderGeneral',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_general.html',1,'mlx::steel::Conv2DInputBlockLoaderGeneral< T, BM, BN, BK, tgp_size, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_general.html#a1d83af561a483432bf8dcb42e734b23b',1,'mlx::steel::Conv2DInputBlockLoaderGeneral::Conv2DInputBlockLoaderGeneral()']]], + ['conv2dinputblockloaderlargefilter_89',['Conv2DInputBlockLoaderLargeFilter',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_large_filter.html',1,'mlx::steel::Conv2DInputBlockLoaderLargeFilter< T, BM, BN, BK, tgp_size, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_large_filter.html#a8755116a535539744e4947bc69f9c50f',1,'mlx::steel::Conv2DInputBlockLoaderLargeFilter::Conv2DInputBlockLoaderLargeFilter()']]], + ['conv2dinputblockloadersmallchannels_90',['Conv2DInputBlockLoaderSmallChannels',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_channels.html',1,'mlx::steel::Conv2DInputBlockLoaderSmallChannels< T, BM, BN, BK, tgp_size, n_channels, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_channels.html#ab9fd3fdeab94470dde3326f1dd5c455a',1,'mlx::steel::Conv2DInputBlockLoaderSmallChannels::Conv2DInputBlockLoaderSmallChannels()']]], + ['conv2dinputblockloadersmallfilter_91',['Conv2DInputBlockLoaderSmallFilter',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_filter.html',1,'mlx::steel::Conv2DInputBlockLoaderSmallFilter< T, BM, BN, BK, tgp_size, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_filter.html#a0a2cbf57c51cd928722e3f06aafcf933',1,'mlx::steel::Conv2DInputBlockLoaderSmallFilter::Conv2DInputBlockLoaderSmallFilter()']]], + ['conv2dweightblockloader_92',['Conv2DWeightBlockLoader',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader.html',1,'mlx::steel::Conv2DWeightBlockLoader< T, BM, BN, BK, tgp_size, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader.html#a9a7dca3512b64cffb6eac305d795831c',1,'mlx::steel::Conv2DWeightBlockLoader::Conv2DWeightBlockLoader()']]], + ['conv2dweightblockloadergeneral_93',['Conv2DWeightBlockLoaderGeneral',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_general.html',1,'mlx::steel::Conv2DWeightBlockLoaderGeneral< T, BM, BN, BK, tgp_size, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_general.html#ad0550fabbdc9297559381a5b488e9af1',1,'mlx::steel::Conv2DWeightBlockLoaderGeneral::Conv2DWeightBlockLoaderGeneral()']]], + ['conv2dweightblockloadersmallchannels_94',['Conv2DWeightBlockLoaderSmallChannels',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_small_channels.html',1,'mlx::steel::Conv2DWeightBlockLoaderSmallChannels< T, BM, BN, BK, tgp_size, n_channels, tgp_padding >'],['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_small_channels.html#ae1806ea1c19713819dee83a38ab35fa6',1,'mlx::steel::Conv2DWeightBlockLoaderSmallChannels::Conv2DWeightBlockLoaderSmallChannels()']]], + ['conv3d_95',['conv3d',['../group__ops.html#ga6e9907d2f14dc4803e4306b3dbc4b3ca',1,'mlx::core']]], + ['conv_5fgeneral_96',['conv_general',['../group__ops.html#ga2236e5dfc7e52e28abf6c21675d0a51e',1,'mlx::core::conv_general(array input, array weight, std::vector< int > stride={}, std::vector< int > padding_lo={}, std::vector< int > padding_hi={}, std::vector< int > kernel_dilation={}, std::vector< int > input_dilation={}, int groups=1, bool flip=false, StreamOrDevice s={})'],['../group__ops.html#gab59f89942cd1efaadffe9e8762e3c99d',1,'mlx::core::conv_general(const array &input, const array &weight, std::vector< int > stride={}, std::vector< int > padding={}, std::vector< int > kernel_dilation={}, std::vector< int > input_dilation={}, int groups=1, bool flip=false, StreamOrDevice s={})']]], + ['conv_5ftranspose1d_97',['conv_transpose1d',['../group__ops.html#gaa30bf1adcd78d1c2595d07b215731714',1,'mlx::core']]], + ['conv_5ftranspose2d_98',['conv_transpose2d',['../group__ops.html#gaebb59971cb9bc45005dc1d398e4f0a3d',1,'mlx::core']]], + ['conv_5ftranspose3d_99',['conv_transpose3d',['../group__ops.html#ga8db814da631d9cd32a8d6563bf4ac530',1,'mlx::core']]], + ['convolution_100',['Convolution',['../classmlx_1_1core_1_1_convolution.html',1,'mlx::core::Convolution'],['../classmlx_1_1core_1_1_convolution.html#a6f1de77b719bb13217b0d8c64cabb8ef',1,'mlx::core::Convolution::Convolution()']]], + ['copy_101',['Copy',['../classmlx_1_1core_1_1_copy.html',1,'mlx::core::Copy'],['../classmlx_1_1core_1_1_copy.html#a6243e044af119105ffaaed7d405cd584',1,'mlx::core::Copy::Copy()']]], + ['copy_102',['copy',['../namespacemlx_1_1core.html#a479648542a2bea151b947b18f0e79dd2',1,'mlx::core::copy()'],['../namespacemlx_1_1core_1_1metal.html#aa215e631e2680f04a591b88d91571719',1,'mlx::core::metal::copy()'],['../group__ops.html#gae306e93af12f774bd80bad6c231b09d6',1,'mlx::core::copy()']]], + ['copy_2eh_103',['copy.h',['../common_2copy_8h.html',1,'(Global Namespace)'],['../metal_2copy_8h.html',1,'(Global Namespace)'],['../metal_2kernels_2copy_8h.html',1,'(Global Namespace)']]], + ['copy_5fg_104',['copy_g',['../metal_2kernels_2copy_8h.html#a778ce2dbfbaa23b24bd5efbe68448c36',1,'copy.h']]], + ['copy_5fg_5fnd1_105',['copy_g_nd1',['../metal_2kernels_2copy_8h.html#aba4530a7db6a61ca36f50e4f5e58fb77',1,'copy.h']]], + ['copy_5fg_5fnd2_106',['copy_g_nd2',['../metal_2kernels_2copy_8h.html#aee678c7c31119f3e609685589f37490c',1,'copy.h']]], + ['copy_5fg_5fnd3_107',['copy_g_nd3',['../metal_2kernels_2copy_8h.html#a821f8f3f3891159a295c66fc25aed1ff',1,'copy.h']]], + ['copy_5fgg_108',['copy_gg',['../metal_2kernels_2copy_8h.html#a1e39c2683eeaf05955e7619fbd34aea5',1,'copy.h']]], + ['copy_5fgg_5fnd1_109',['copy_gg_nd1',['../metal_2kernels_2copy_8h.html#a3278d9c999718bee3ccbe2922f501bf1',1,'copy.h']]], + ['copy_5fgg_5fnd2_110',['copy_gg_nd2',['../metal_2kernels_2copy_8h.html#a3e2d3cc7f34f56170409b6735f51a950',1,'copy.h']]], + ['copy_5fgg_5fnd3_111',['copy_gg_nd3',['../metal_2kernels_2copy_8h.html#a59f43b5bffed936d7559ceb06a10aabd',1,'copy.h']]], + ['copy_5fgpu_112',['copy_gpu',['../namespacemlx_1_1core.html#addaa46a13ac2deb1d9ce621338320e0e',1,'mlx::core::copy_gpu(const array &src, array &out, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#a6a6f4e46c8fc44fdc74c50ace02bcf38',1,'mlx::core::copy_gpu(const array &src, array &out, CopyType ctype)']]], + ['copy_5fgpu_5finplace_113',['copy_gpu_inplace',['../namespacemlx_1_1core.html#a69e30f5d30a6d72ac0ffe4886f24b7ba',1,'mlx::core::copy_gpu_inplace(const array &in, array &out, const std::vector< int > &data_shape, const std::vector< stride_t > &i_strides, const std::vector< stride_t > &o_strides, int64_t i_offset, int64_t o_offset, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#a8e1ccb0ed9387b0a789311d9f8964803',1,'mlx::core::copy_gpu_inplace(const array &src, array &out, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#ae55b801b09ccf55cba96278163a9b1ef',1,'mlx::core::copy_gpu_inplace(const array &in, array &out, const std::vector< int64_t > &istride, int64_t ioffset, CopyType ctype, const Stream &s)']]], + ['copy_5fhartley_114',['copy_hartley',['../namespacepocketfft_1_1detail.html#abac3fcc8ce83800d228774f64c28d4c3',1,'pocketfft::detail::copy_hartley(const multi_iter< vlen > &it, const vtype_t< T > *src, ndarr< T > &dst)'],['../namespacepocketfft_1_1detail.html#ae7b44d2773d9d06a9787aff01d66b3ed',1,'pocketfft::detail::copy_hartley(const multi_iter< vlen > &it, const T *src, ndarr< T > &dst)']]], + ['copy_5finplace_115',['copy_inplace',['../namespacemlx_1_1core.html#a98495894a796b2cc6d022e7a03432c64',1,'mlx::core::copy_inplace(const array &src, array &dst, CopyType ctype)'],['../namespacemlx_1_1core.html#aad636e2d0b2f882cadd1b438f4daa9ed',1,'mlx::core::copy_inplace(const array &src, array &dst, const std::vector< int > &data_shape, const std::vector< stride_t > &i_strides, const std::vector< stride_t > &o_strides, int64_t i_offset, int64_t o_offset, CopyType ctype)']]], + ['copy_5finput_116',['copy_input',['../namespacepocketfft_1_1detail.html#aff05be3064743c1143b19318ab12ad4a',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< cmplx< T > > &src, cmplx< vtype_t< T > > *dst)'],['../namespacepocketfft_1_1detail.html#a30fc708f9d8f9cfa74194925c7863c0a',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< T > &src, vtype_t< T > *dst)'],['../namespacepocketfft_1_1detail.html#a3387bd35f237870e42b8461769e6aec4',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< T > &src, T *dst)']]], + ['copy_5foutput_117',['copy_output',['../namespacepocketfft_1_1detail.html#a1523a037300a8da05db210b802d9cb0e',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const cmplx< vtype_t< T > > *src, ndarr< cmplx< T > > &dst)'],['../namespacepocketfft_1_1detail.html#a21980853aca4d92ed06e3dcffe7ef660',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const vtype_t< T > *src, ndarr< T > &dst)'],['../namespacepocketfft_1_1detail.html#a310481c334e46674710ba794ad7403c0',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const T *src, ndarr< T > &dst)']]], + ['copy_5fs_118',['copy_s',['../metal_2kernels_2copy_8h.html#aef09f9b9475345b1bba121d037d222ea',1,'copy.h']]], + ['copy_5fs2_119',['copy_s2',['../metal_2kernels_2copy_8h.html#a8023e9335cc5334847a8d315042be3a3',1,'copy.h']]], + ['copy_5fshared_5fbuffer_120',['copy_shared_buffer',['../classmlx_1_1core_1_1array.html#a28df7a333d90a311c49bc4bce7a1ad6d',1,'mlx::core::array::copy_shared_buffer(const array &other, const std::vector< size_t > &strides, Flags flags, size_t data_size, size_t offset=0)'],['../classmlx_1_1core_1_1array.html#a92974c656c35a972ad241f80584bbd29',1,'mlx::core::array::copy_shared_buffer(const array &other)']]], + ['copy_5fv_121',['copy_v',['../metal_2kernels_2copy_8h.html#ae26a13e0c8e6c15f7b10078e65970659',1,'copy.h']]], + ['copy_5fv2_122',['copy_v2',['../metal_2kernels_2copy_8h.html#aee14a5326f53d9b30b0b38e27d180ef3',1,'copy.h']]], + ['copytype_123',['CopyType',['../namespacemlx_1_1core.html#abd84ff6c5245e4e170b2ef5247594337',1,'mlx::core']]], + ['core_20array_20operations_124',['Core array operations',['../group__ops.html',1,'']]], + ['cos_125',['Cos',['../struct_cos.html',1,'Cos'],['../classmlx_1_1core_1_1_cos.html',1,'mlx::core::Cos'],['../structmlx_1_1core_1_1detail_1_1_cos.html',1,'mlx::core::detail::Cos'],['../classmlx_1_1core_1_1_cos.html#a2acb9fcf0901462189c476756fd99995',1,'mlx::core::Cos::Cos()']]], + ['cos_126',['cos',['../namespacepocketfft_1_1detail.html#a499c1e8b7d79a5272af024f46c63ff9d',1,'pocketfft::detail::cos()'],['../namespacemetal.html#a2fa4778a6fe2fa43253ea724e5a608a3',1,'metal::cos()'],['../namespacemetal_1_1fast.html#a75b6bb32fa3870eda46a7bfc9f481f88',1,'metal::fast::cos()'],['../namespacemetal_1_1precise.html#ac4941f62e7d8ab9d7cabbd967aa9f220',1,'metal::precise::cos()'],['../group__ops.html#ga39dfdf72b556012aa35ff27a94116e74',1,'mlx::core::cos()']]], + ['cosh_127',['Cosh',['../struct_cosh.html',1,'Cosh'],['../classmlx_1_1core_1_1_cosh.html',1,'mlx::core::Cosh'],['../structmlx_1_1core_1_1detail_1_1_cosh.html',1,'mlx::core::detail::Cosh'],['../classmlx_1_1core_1_1_cosh.html#a44e8ac2e09a55ec32e9dc6641eedc8f1',1,'mlx::core::Cosh::Cosh()']]], + ['cosh_128',['cosh',['../namespacemetal.html#a8a68a88cc110830d057dbd71431b93c0',1,'metal::cosh()'],['../namespacemetal_1_1fast.html#a31544ad9de28012a4ddda86e3966a77e',1,'metal::fast::cosh()'],['../namespacemetal_1_1precise.html#a72d86d508300a9b58f4ccbbe70da4fbc',1,'metal::precise::cosh()'],['../group__ops.html#ga2181b71cda88007a3092be4795ff0715',1,'mlx::core::cosh()']]], + ['cosine_129',['cosine',['../structpocketfft_1_1detail_1_1_exec_dcst.html#a185023fc1e386cc8f233b79c49c1fd8a',1,'pocketfft::detail::ExecDcst']]], + ['cospi_130',['cospi',['../namespacemetal.html#a5c2f37939ad705ddea4409d3bedb8ce1',1,'metal::cospi()'],['../namespacemetal_1_1fast.html#a9906b41f75319b384ffb570cc94d67ce',1,'metal::fast::cospi()'],['../namespacemetal_1_1precise.html#a2392b78bd196efdbbac65901c4ab20e7',1,'metal::precise::cospi()']]], + ['cost_5fguess_131',['cost_guess',['../structpocketfft_1_1detail_1_1util.html#ad3d874bc3fb0048df2270779a15d4bd0',1,'pocketfft::detail::util']]], + ['count_5fdown_132',['count_down',['../classpocketfft_1_1detail_1_1threading_1_1latch.html#a81d6597189b40410e35f3cd653fd1342',1,'pocketfft::detail::threading::latch']]], + ['cpu_133',['cpu',['../structmlx_1_1core_1_1_device.html#a69ee81924251dec96f1945c9d91506fd',1,'mlx::core::Device::cpu'],['../structmlx_1_1core_1_1_device.html#ac45b3de9b3458d8f31005136cde20fdbad9747e2da342bdb995f6389533ad1a3d',1,'mlx::core::Device::cpu']]], + ['cross_134',['cross',['../namespacemlx_1_1core_1_1linalg.html#abcda3fbda45183c21e7f27aa0dde64e6',1,'mlx::core::linalg']]], + ['ctile_135',['Ctile',['../structmlx_1_1steel_1_1_block_m_m_a.html#a81838da5d81e62d372d581be599c5a88',1,'mlx::steel::BlockMMA']]], + ['cummax_136',['CumMax',['../struct_cum_max.html',1,'']]], + ['cummax_137',['cummax',['../group__ops.html#gaee37cac8476e8f8d666bcded5bc59143',1,'mlx::core']]], + ['cummin_138',['CumMin',['../struct_cum_min.html',1,'']]], + ['cummin_139',['cummin',['../group__ops.html#ga19c1bf6929fe8d66b9cd408946aea6a8',1,'mlx::core']]], + ['cumprod_140',['CumProd',['../struct_cum_prod.html',1,'']]], + ['cumprod_141',['cumprod',['../group__ops.html#ga0d71dfbc14ef3ed564b0c5ee26af680f',1,'mlx::core']]], + ['cumprod_3c_20bool_20_3e_142',['CumProd< bool >',['../struct_cum_prod_3_01bool_01_4.html',1,'']]], + ['cumsum_143',['CumSum',['../struct_cum_sum.html',1,'']]], + ['cumsum_144',['cumsum',['../group__ops.html#gaddc825a5c173e195ab0fda83ad630420',1,'mlx::core']]], + ['custom_145',['Custom',['../classmlx_1_1core_1_1fast_1_1_custom.html',1,'mlx::core::fast::Custom'],['../classmlx_1_1core_1_1fast_1_1_custom.html#a4186fea23f7156c38960426821fca313',1,'mlx::core::fast::Custom::Custom()']]], + ['custom_5ffunction_146',['custom_function',['../namespacemlx_1_1core.html#a8d3ca5fbaecdb995660c24cde5aeebaf',1,'mlx::core']]], + ['custom_5fvjp_147',['custom_vjp',['../namespacemlx_1_1core.html#a9290596250fa308df4c69b44483bb8aa',1,'mlx::core']]], + ['customkernel_148',['CustomKernel',['../classmlx_1_1core_1_1fast_1_1_custom_kernel.html',1,'mlx::core::fast::CustomKernel'],['../classmlx_1_1core_1_1fast_1_1_custom_kernel.html#a954893e07f0d36715b4e1e414b6f2153',1,'mlx::core::fast::CustomKernel::CustomKernel()']]], + ['customkernelshapeinfo_149',['CustomKernelShapeInfo',['../structmlx_1_1core_1_1fast_1_1_custom_kernel_shape_info.html',1,'mlx::core::fast']]], + ['customtransforms_150',['CustomTransforms',['../classmlx_1_1core_1_1_custom_transforms.html',1,'mlx::core::CustomTransforms'],['../classmlx_1_1core_1_1_custom_transforms.html#ab52abadb9c6f6db83d087c7b751be488',1,'mlx::core::CustomTransforms::CustomTransforms()']]] ]; diff --git a/docs/build/html/search/all_7.js b/docs/build/html/search/all_7.js index d31b9bda7..50f643bac 100644 --- a/docs/build/html/search/all_7.js +++ b/docs/build/html/search/all_7.js @@ -67,7 +67,7 @@ var searchData= ['get_5fpool_64',['get_pool',['../namespacepocketfft_1_1detail_1_1threading.html#a7ec2b3f99232bd0f15f7b022c59d139a',1,'pocketfft::detail::threading']]], ['get_5fprimitive_5fstring_65',['get_primitive_string',['../namespacemlx_1_1core.html#ad4be35b310a252edd80d9cf04f094a60',1,'mlx::core']]], ['get_5fquantized_5fkernel_66',['get_quantized_kernel',['../namespacemlx_1_1core.html#aa3faeae5378bfaafe3ce3432a051e43e',1,'mlx::core']]], - ['get_5freduce_5finit_5fkernel_67',['get_reduce_init_kernel',['../namespacemlx_1_1core.html#a51c4bb09230348bd0252e22bfdc9bc89',1,'mlx::core']]], + ['get_5freduce_5finit_5fkernel_67',['get_reduce_init_kernel',['../namespacemlx_1_1core.html#a3bd386cb6db09f636963ce66ceaf8647',1,'mlx::core']]], ['get_5freduce_5fkernel_68',['get_reduce_kernel',['../namespacemlx_1_1core.html#a7aa91fcfe8b9caa42d60a957f11bfe6b',1,'mlx::core']]], ['get_5freduction_5fplan_69',['get_reduction_plan',['../namespacemlx_1_1core.html#ac97b5a6f009ca3d99854ce9512c20dba',1,'mlx::core']]], ['get_5fscan_5fkernel_70',['get_scan_kernel',['../namespacemlx_1_1core.html#aeefaff208444d3fa61ecc0946fe1de5f',1,'mlx::core']]], diff --git a/docs/build/html/search/all_d.js b/docs/build/html/search/all_d.js index b941fdcba..b1b018f5f 100644 --- a/docs/build/html/search/all_d.js +++ b/docs/build/html/search/all_d.js @@ -29,84 +29,85 @@ var searchData= ['max_5fthreads_26',['max_threads',['../namespacepocketfft_1_1detail_1_1threading.html#a2d5c0729f0b66cf061918baea4337d70',1,'pocketfft::detail::threading']]], ['maximum_27',['Maximum',['../struct_maximum.html',1,'Maximum'],['../structmlx_1_1core_1_1detail_1_1_maximum.html',1,'mlx::core::detail::Maximum'],['../classmlx_1_1core_1_1_maximum.html',1,'mlx::core::Maximum'],['../classmlx_1_1core_1_1_maximum.html#a28389307e385efe1b2955b86b115e816',1,'mlx::core::Maximum::Maximum()']]], ['maximum_28',['maximum',['../group__ops.html#ga7ade2ea305e2e4219c3609443fb5db8d',1,'mlx::core']]], - ['mb_5fblock_5fmerge_29',['mb_block_merge',['../sort_8h.html#ab381cd57f344bc7304ab580bfdc78807',1,'sort.h']]], - ['mb_5fblock_5fpartition_30',['mb_block_partition',['../sort_8h.html#a32cbe4163b8b0f5cb2c97b256119a4b2',1,'sort.h']]], - ['mb_5fblock_5fsort_31',['mb_block_sort',['../sort_8h.html#aa48ff1aff1e9dc1301b6781aa0721d6b',1,'sort.h']]], - ['mean_32',['mean',['../group__ops.html#gade46e768fd46b8b640eb16f26abeecef',1,'mlx::core::mean(const array &a, bool keepdims, StreamOrDevice s={})'],['../group__ops.html#ga52b59fdd8e8430538e564f5bbcfa31e6',1,'mlx::core::mean(const array &a, StreamOrDevice s={})'],['../group__ops.html#ga066161f3d3e395a1d76c638cb680d444',1,'mlx::core::mean(const array &a, const std::vector< int > &axes, bool keepdims=false, StreamOrDevice s={})'],['../group__ops.html#ga45fba73eab0e3b6e128ed3ce2f43a5da',1,'mlx::core::mean(const array &a, int axis, bool keepdims=false, StreamOrDevice s={})']]], - ['median3_33',['median3',['../namespacemetal.html#aa3ff49457ce3c93fc1c0897fd1525157',1,'metal::median3()'],['../namespacemetal_1_1fast.html#a742b55f1e4369921ee7f60d70185bfbc',1,'metal::fast::median3()'],['../namespacemetal_1_1precise.html#a14555ff99c4388493fec48e070144ae2',1,'metal::precise::median3()']]], - ['merge_5fpartition_34',['merge_partition',['../struct_block_merge_sort.html#ab2300cbecb23f3433bad888924c831ca',1,'BlockMergeSort::merge_partition()'],['../struct_kernel_multi_block_merge_sort.html#ab15895b4233aba0e279cc44a07a201fe',1,'KernelMultiBlockMergeSort::merge_partition()']]], - ['merge_5fstep_35',['merge_step',['../struct_block_merge_sort.html#ab65f190edf1851b37c39ad49ce99a43c',1,'BlockMergeSort']]], - ['meshgrid_36',['meshgrid',['../group__ops.html#ga577c911618575314de63d1060656a26e',1,'mlx::core']]], - ['metal_37',['metal',['../namespacemetal.html',1,'']]], - ['metal_2eh_38',['metal.h',['../metal_8h.html',1,'']]], - ['metal_3a_3afast_39',['fast',['../namespacemetal_1_1fast.html',1,'metal']]], - ['metal_3a_3aprecise_40',['precise',['../namespacemetal_1_1precise.html',1,'metal']]], - ['metal_5fimpl_2eh_41',['metal_impl.h',['../metal__impl_8h.html',1,'']]], - ['metal_5fkernel_42',['metal_kernel',['../namespacemlx_1_1core_1_1fast.html#ab16436b465dc10ce472193d541d8426e',1,'mlx::core::fast']]], - ['metalallocator_43',['MetalAllocator',['../classmlx_1_1core_1_1metal_1_1_metal_allocator.html',1,'mlx::core::metal']]], - ['metalkernelfunction_44',['MetalKernelFunction',['../namespacemlx_1_1core_1_1fast.html#a0e8c2c4ea7a946568c8fe5b4810417e0',1,'mlx::core::fast']]], - ['min_45',['Min',['../struct_min.html',1,'Min< U >'],['../classmlx_1_1core_1_1distributed_1_1_all_reduce.html#abb4560980e5d01aed14175ce8f6fc924a4f685dcd48e6614d6bb2ccda4f2686ef',1,'mlx::core::distributed::AllReduce::Min'],['../classmlx_1_1core_1_1_reduce.html#a0848518b16ae6d4043d6be247bdf31c9a0d3d1f5c94725bdc42fa692e2c074418',1,'mlx::core::Reduce::Min'],['../classmlx_1_1core_1_1_scan.html#a47bf2ec54ead4b8f00f9f188518630f1a7d2ee8f14f2e70a9d47170fecc6da898',1,'mlx::core::Scan::Min'],['../classmlx_1_1core_1_1_scatter.html#a614d19af11dc30644b2b4941033b613cad914e4c3475ce9858f2de4bf35dcfdbf',1,'mlx::core::Scatter::Min']]], - ['min_46',['min',['../struct_limits.html#a6e81584ba65a4dc6ff9366b458e3a20e',1,'Limits::min'],['../struct_limits_3_01uint8__t_01_4.html#a408bd5a337e7292f06e63da81193629a',1,'Limits< uint8_t >::min'],['../struct_limits_3_01uint16__t_01_4.html#ae173984c3be8b6750f27daed581805fe',1,'Limits< uint16_t >::min'],['../struct_limits_3_01uint32__t_01_4.html#ab0c3975e02053b234c7b606ababa66e1',1,'Limits< uint32_t >::min'],['../struct_limits_3_01uint64__t_01_4.html#a80627f39e951398283942cefa48f4dd0',1,'Limits< uint64_t >::min'],['../struct_limits_3_01int8__t_01_4.html#a7a809307d2bba80382f0645d277eaa4b',1,'Limits< int8_t >::min'],['../struct_limits_3_01int16__t_01_4.html#adca7139647801e223c35b0abc7da5240',1,'Limits< int16_t >::min'],['../struct_limits_3_01int32__t_01_4.html#af336a1b22a8ed6a83a4cfb5bf8869771',1,'Limits< int32_t >::min'],['../struct_limits_3_01int64__t_01_4.html#a1c90fb96af515badaccaa835b08f7428',1,'Limits< int64_t >::min'],['../struct_limits_3_01half_01_4.html#aca7b036c257878bf1b80912fb5d4516d',1,'Limits< half >::min'],['../struct_limits_3_01float_01_4.html#a3225e334d372ee86128c89a440d8648f',1,'Limits< float >::min'],['../struct_limits_3_01bfloat16__t_01_4.html#a2fd1811b9f615b2b897904bc27d1cb49',1,'Limits< bfloat16_t >::min'],['../struct_limits_3_01bool_01_4.html#a139f787b57536d455490b8ef801d37cc',1,'Limits< bool >::min'],['../struct_limits_3_01complex64__t_01_4.html#aa67b04aa7abcd67f7af0808737ab8e14',1,'Limits< complex64_t >::min'],['../structmetal_1_1__numeric__limits__impl_3_01bfloat16__t_01_4.html#adaed80031f5ca0ff69d30ec4c5d0c98f',1,'metal::_numeric_limits_impl< bfloat16_t >::min()'],['../namespacemetal.html#a6653b28c9473087141eddce39878d4d3',1,'metal::min()'],['../namespacemetal_1_1fast.html#a3e958e56a4712687c381a0b64d123e61',1,'metal::fast::min()'],['../namespacemetal_1_1precise.html#afed0da2f7df3505b5dffa2389c3cb36e',1,'metal::precise::min()'],['../group__ops.html#gab27599802617a4c8f9964ab5f4ffee12',1,'mlx::core::min(const array &a, bool keepdims, StreamOrDevice s={})'],['../group__ops.html#ga0140b91e9cdfc3fef0da8e332f65a9e8',1,'mlx::core::min(const array &a, StreamOrDevice s={})'],['../group__ops.html#ga6efb83cd46436678c8f8c4af15cc00f5',1,'mlx::core::min(const array &a, const std::vector< int > &axes, bool keepdims=false, StreamOrDevice s={})'],['../group__ops.html#ga36fa315eef677f4143868f552cd26d03',1,'mlx::core::min(const array &a, int axis, bool keepdims=false, StreamOrDevice s={})']]], - ['min3_47',['min3',['../namespacemetal.html#a005510c8c0f964ce2b8aad3ba76a7a3f',1,'metal::min3()'],['../namespacemetal_1_1fast.html#a606a4c1b34ce05ea89ca5af81724036f',1,'metal::fast::min3()'],['../namespacemetal_1_1precise.html#a4d37ce31c3549ca4772a4ee29798e231',1,'metal::precise::min3()']]], - ['min_5fexponent_48',['min_exponent',['../structmetal_1_1__numeric__limits__impl_3_01bfloat16__t_01_4.html#a13829f8c7a7c0efdc8946eff5d3c9470',1,'metal::_numeric_limits_impl< bfloat16_t >']]], - ['min_5fexponent10_49',['min_exponent10',['../structmetal_1_1__numeric__limits__impl_3_01bfloat16__t_01_4.html#aeaed172780720e06b8731cef3177e277',1,'metal::_numeric_limits_impl< bfloat16_t >']]], - ['minimum_50',['Minimum',['../struct_minimum.html',1,'Minimum'],['../structmlx_1_1core_1_1detail_1_1_minimum.html',1,'mlx::core::detail::Minimum'],['../classmlx_1_1core_1_1_minimum.html',1,'mlx::core::Minimum'],['../classmlx_1_1core_1_1_minimum.html#ab0f2ce17108df44b82cff68886b0f6f5',1,'mlx::core::Minimum::Minimum()']]], - ['minimum_51',['minimum',['../group__ops.html#ga49ba00c090f81f331c91b0c97040bce0',1,'mlx::core']]], - ['mlx_52',['mlx',['../namespacemlx.html',1,'']]], - ['mlx_2eh_53',['mlx.h',['../mlx_8h.html',1,'']]], - ['mlx_3a_3acore_54',['core',['../namespacemlx_1_1core.html',1,'mlx']]], - ['mlx_3a_3acore_3a_3aallocator_55',['allocator',['../namespacemlx_1_1core_1_1allocator.html',1,'mlx::core']]], - ['mlx_3a_3acore_3a_3adetail_56',['detail',['../namespacemlx_1_1core_1_1detail.html',1,'mlx::core']]], - ['mlx_3a_3acore_3a_3adistributed_57',['distributed',['../namespacemlx_1_1core_1_1distributed.html',1,'mlx::core']]], - ['mlx_3a_3acore_3a_3adistributed_3a_3adetail_58',['detail',['../namespacemlx_1_1core_1_1distributed_1_1detail.html',1,'mlx::core::distributed']]], - ['mlx_3a_3acore_3a_3afast_59',['fast',['../namespacemlx_1_1core_1_1fast.html',1,'mlx::core']]], - ['mlx_3a_3acore_3a_3afft_60',['fft',['../namespacemlx_1_1core_1_1fft.html',1,'mlx::core']]], - ['mlx_3a_3acore_3a_3aio_61',['io',['../namespacemlx_1_1core_1_1io.html',1,'mlx::core']]], - ['mlx_3a_3acore_3a_3alinalg_62',['linalg',['../namespacemlx_1_1core_1_1linalg.html',1,'mlx::core']]], - ['mlx_3a_3acore_3a_3ametal_63',['metal',['../namespacemlx_1_1core_1_1metal.html',1,'mlx::core']]], - ['mlx_3a_3acore_3a_3arandom_64',['random',['../namespacemlx_1_1core_1_1random.html',1,'mlx::core']]], - ['mlx_3a_3acore_3a_3ascheduler_65',['scheduler',['../namespacemlx_1_1core_1_1scheduler.html',1,'mlx::core']]], - ['mlx_3a_3asteel_66',['steel',['../namespacemlx_1_1steel.html',1,'mlx']]], - ['mlx_5fatomic_67',['mlx_atomic',['../structmlx__atomic.html',1,'']]], - ['mlx_5fatomic_3c_20t_2c_20enable_5fif_5ft_3c_20is_5fmetal_5fatomic_3c_20t_20_3e_20_3e_20_3e_68',['mlx_atomic< T, enable_if_t< is_metal_atomic< T > > >',['../structmlx__atomic_3_01_t_00_01enable__if__t_3_01is__metal__atomic_3_01_t_01_4_01_4_01_4.html',1,'']]], - ['mlx_5fatomic_5fcompare_5fexchange_5fweak_5fexplicit_69',['mlx_atomic_compare_exchange_weak_explicit',['../atomic_8h.html#ad7f32327ff66354cfa2f0cfdac79316f',1,'mlx_atomic_compare_exchange_weak_explicit(device mlx_atomic< T > *object, thread T *expected, T val, size_t offset): atomic.h'],['../atomic_8h.html#aa8f47b2e9b95d4b00ad51f08b070deb5',1,'mlx_atomic_compare_exchange_weak_explicit(device mlx_atomic< T > *object, thread uint *expected, uint val, size_t offset): atomic.h']]], - ['mlx_5fatomic_5ffetch_5fadd_5fexplicit_70',['mlx_atomic_fetch_add_explicit',['../atomic_8h.html#aad448d9e06e001700b65ca8317216a3b',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fand_5fexplicit_71',['mlx_atomic_fetch_and_explicit',['../atomic_8h.html#a253e3c870c0ddc7c28ab2f6ca2c3eae5',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fmax_5fexplicit_72',['mlx_atomic_fetch_max_explicit',['../atomic_8h.html#ac480f2b459a8ad9095cee353e152d00c',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fmax_5fexplicit_3c_20float_20_3e_73',['mlx_atomic_fetch_max_explicit< float >',['../atomic_8h.html#a1dce2abfa16417122c4d2bf261129ae4',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fmin_5fexplicit_74',['mlx_atomic_fetch_min_explicit',['../atomic_8h.html#a2ec33dca0039bd944d73d1c2b378cc19',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fmin_5fexplicit_3c_20float_20_3e_75',['mlx_atomic_fetch_min_explicit< float >',['../atomic_8h.html#ab7d1dc49f319f239b7ee0b7c72976dd0',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fmul_5fexplicit_76',['mlx_atomic_fetch_mul_explicit',['../atomic_8h.html#adfdbea60436f14f1af9ce36e2a0a77a3',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5for_5fexplicit_77',['mlx_atomic_fetch_or_explicit',['../atomic_8h.html#ab7391f197001471e4788312bdb6ab37a',1,'atomic.h']]], - ['mlx_5fatomic_5fload_5fexplicit_78',['mlx_atomic_load_explicit',['../atomic_8h.html#a253a4e8c2c5768a069e2791b627dfc99',1,'atomic.h']]], - ['mlx_5fatomic_5fstore_5fexplicit_79',['mlx_atomic_store_explicit',['../atomic_8h.html#a0ae453140b0819a4c02f265334de98c0',1,'atomic.h']]], - ['mlx_5flapack_5ffunc_80',['MLX_LAPACK_FUNC',['../lapack_8h.html#ae22db9704827bf013a0a61f21a47464b',1,'lapack.h']]], - ['mlx_5fmtl_5fconst_81',['MLX_MTL_CONST',['../kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd',1,'MLX_MTL_CONST: gemv_masked.h'],['../quantized_8h.html#a0386011c52d03e60885a31e6fbd903dd',1,'MLX_MTL_CONST: quantized.h'],['../sort_8h.html#a0386011c52d03e60885a31e6fbd903dd',1,'MLX_MTL_CONST: sort.h']]], - ['mlx_5fmtl_5floop_5funroll_82',['MLX_MTL_LOOP_UNROLL',['../sort_8h.html#ad34b622323cebef136669fedd7229515',1,'sort.h']]], - ['mlx_5fmtl_5fpragma_5funroll_83',['MLX_MTL_PRAGMA_UNROLL',['../kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3',1,'MLX_MTL_PRAGMA_UNROLL: gemv_masked.h'],['../backend_2metal_2kernels_2utils_8h.html#a069b682d7d21827461544817d722bfd3',1,'MLX_MTL_PRAGMA_UNROLL: utils.h']]], - ['mlxconvparams_84',['MLXConvParams',['../struct_m_l_x_conv_params.html',1,'']]], - ['mlxconvparams_3c_202_20_3e_85',['MLXConvParams< 2 >',['../struct_m_l_x_conv_params.html',1,'']]], - ['mlxfastattentionparams_86',['MLXFastAttentionParams',['../struct_m_l_x_fast_attention_params.html',1,'']]], - ['mlxscaleddotproductattentionparams_87',['MLXScaledDotProductAttentionParams',['../struct_m_l_x_scaled_dot_product_attention_params.html',1,'']]], - ['mma_88',['mma',['../structmlx_1_1steel_1_1_base_m_m_a_frag_3_01_t_00_018_00_018_01_4.html#a8028512f5a3d2b6acaf966be529627a3',1,'mlx::steel::BaseMMAFrag< T, 8, 8 >::mma(thread frag_type &D, thread frag_type &A, thread frag_type &B, thread frag_type &C)'],['../structmlx_1_1steel_1_1_base_m_m_a_frag_3_01_t_00_018_00_018_01_4.html#a1868f57d57c8adedab2c58492ec76946',1,'mlx::steel::BaseMMAFrag< T, 8, 8 >::mma(thread mat_type &D, thread mat_type &A, thread mat_type &B, thread mat_type &C)'],['../structmlx_1_1steel_1_1_block_m_m_a.html#a6a2c2a6d5e767d52c41b42a9d36086b0',1,'mlx::steel::BlockMMA::mma()']]], - ['mma_2eh_89',['mma.h',['../mma_8h.html',1,'']]], - ['mma_5ft_90',['mma_t',['../structmlx_1_1steel_1_1_g_e_m_m_kernel.html#add8c6a31011a4895667c2a94a5af3782',1,'mlx::steel::GEMMKernel']]], - ['mmafrag_5facc_5ft_91',['MMAFrag_acc_t',['../structmlx_1_1steel_1_1_block_m_m_a.html#ae2c42cb6d0dde785859164c195f4d13c',1,'mlx::steel::BlockMMA']]], - ['mmafrag_5ft_92',['MMAFrag_t',['../structmlx_1_1steel_1_1_m_m_a_tile.html#abe33de70e34300745bad9aa822fd0382',1,'mlx::steel::MMATile']]], - ['mmatile_93',['MMATile',['../structmlx_1_1steel_1_1_m_m_a_tile.html',1,'mlx::steel::MMATile< T, kTileRows_, kTileCols_, MMAFrag_ >'],['../structmlx_1_1steel_1_1_m_m_a_tile.html#aa3fb310dd08ec23c334511f7b316d1b6',1,'mlx::steel::MMATile::MMATile()']]], - ['mmatile_3c_20float_2c_201_2c_20tn_2c_20mlx_3a_3asteel_3a_3abasemmafrag_20_3e_94',['MMATile< float, 1, TN, mlx::steel::BaseMMAFrag >',['../structmlx_1_1steel_1_1_m_m_a_tile.html',1,'mlx::steel']]], - ['mmatile_3c_20float_2c_20tm_2c_201_2c_20mlx_3a_3asteel_3a_3abasemmafrag_20_3e_95',['MMATile< float, TM, 1, mlx::steel::BaseMMAFrag >',['../structmlx_1_1steel_1_1_m_m_a_tile.html',1,'mlx::steel']]], - ['mmatile_3c_20float_2c_20tm_2c_20tn_2c_20mlx_3a_3asteel_3a_3abasemmafrag_20_3e_96',['MMATile< float, TM, TN, mlx::steel::BaseMMAFrag >',['../structmlx_1_1steel_1_1_m_m_a_tile.html',1,'mlx::steel']]], - ['move_5fshared_5fbuffer_97',['move_shared_buffer',['../classmlx_1_1core_1_1array.html#acce00db63e0f3d80f797b02397ade836',1,'mlx::core::array::move_shared_buffer(array other, const std::vector< size_t > &strides, Flags flags, size_t data_size, size_t offset=0)'],['../classmlx_1_1core_1_1array.html#a38d7ad605f8282e5e49d0c09e0555c78',1,'mlx::core::array::move_shared_buffer(array other)']]], - ['moveaxis_98',['moveaxis',['../group__ops.html#ga24067d10a842db2c9d509ea48135a2c3',1,'mlx::core']]], - ['mpinplace_99',['MPINPLACE',['../namespacepocketfft_1_1detail.html#af5eedf3cdfc83c0a30807092c39a9ce2',1,'pocketfft::detail']]], - ['mtl_5fconst_100',['MTL_CONST',['../defines_8h.html#a767ed9f2604de22b259cee02c4ce1d22',1,'defines.h']]], - ['mtl_5fdevice_101',['mtl_device',['../classmlx_1_1core_1_1metal_1_1_device.html#a31dba377f2be44a746db10d1b9367653',1,'mlx::core::metal::Device']]], - ['mtl_5fresidency_5fset_102',['mtl_residency_set',['../classmlx_1_1core_1_1metal_1_1_residency_set.html#ac4bfe5ef5e2eaebc458a1ed1953d15e9',1,'mlx::core::metal::ResidencySet']]], - ['mtlfclist_103',['MTLFCList',['../namespacemlx_1_1core_1_1metal.html#a616e09a1ef321d527770721cef264c54',1,'mlx::core::metal']]], - ['mtx_104',['mtx',['../structmlx_1_1core_1_1scheduler_1_1_stream_thread.html#a70410c9e612f871663929f1e8441a976',1,'mlx::core::scheduler::StreamThread']]], - ['multi_5fiter_105',['multi_iter',['../classpocketfft_1_1detail_1_1multi__iter.html',1,'pocketfft::detail::multi_iter< N >'],['../classpocketfft_1_1detail_1_1multi__iter.html#a9be43bb18840202da6d17988fccc64b9',1,'pocketfft::detail::multi_iter::multi_iter()']]], - ['multiply_106',['Multiply',['../structmlx_1_1core_1_1detail_1_1_multiply.html',1,'mlx::core::detail::Multiply'],['../classmlx_1_1core_1_1_multiply.html',1,'mlx::core::Multiply'],['../struct_multiply.html',1,'Multiply'],['../classmlx_1_1core_1_1_multiply.html#aca5c50f900321f3eb4d6fbcbc225c00c',1,'mlx::core::Multiply::Multiply()']]], - ['multiply_107',['multiply',['../group__ops.html#gaf57392e641640b5d06e4c99518391c38',1,'mlx::core']]], - ['multivariate_5fnormal_108',['multivariate_normal',['../namespacemlx_1_1core_1_1random.html#a8c37da3c1c0c561cad7499d6d9db81fb',1,'mlx::core::random']]] + ['maybeinsertbarrier_29',['maybeInsertBarrier',['../structmlx_1_1core_1_1metal_1_1_command_encoder.html#ad538ae88f90560063f9ba502e2795991',1,'mlx::core::metal::CommandEncoder']]], + ['mb_5fblock_5fmerge_30',['mb_block_merge',['../sort_8h.html#ab381cd57f344bc7304ab580bfdc78807',1,'sort.h']]], + ['mb_5fblock_5fpartition_31',['mb_block_partition',['../sort_8h.html#a32cbe4163b8b0f5cb2c97b256119a4b2',1,'sort.h']]], + ['mb_5fblock_5fsort_32',['mb_block_sort',['../sort_8h.html#aa48ff1aff1e9dc1301b6781aa0721d6b',1,'sort.h']]], + ['mean_33',['mean',['../group__ops.html#gade46e768fd46b8b640eb16f26abeecef',1,'mlx::core::mean(const array &a, bool keepdims, StreamOrDevice s={})'],['../group__ops.html#ga52b59fdd8e8430538e564f5bbcfa31e6',1,'mlx::core::mean(const array &a, StreamOrDevice s={})'],['../group__ops.html#ga066161f3d3e395a1d76c638cb680d444',1,'mlx::core::mean(const array &a, const std::vector< int > &axes, bool keepdims=false, StreamOrDevice s={})'],['../group__ops.html#ga45fba73eab0e3b6e128ed3ce2f43a5da',1,'mlx::core::mean(const array &a, int axis, bool keepdims=false, StreamOrDevice s={})']]], + ['median3_34',['median3',['../namespacemetal.html#aa3ff49457ce3c93fc1c0897fd1525157',1,'metal::median3()'],['../namespacemetal_1_1fast.html#a742b55f1e4369921ee7f60d70185bfbc',1,'metal::fast::median3()'],['../namespacemetal_1_1precise.html#a14555ff99c4388493fec48e070144ae2',1,'metal::precise::median3()']]], + ['merge_5fpartition_35',['merge_partition',['../struct_block_merge_sort.html#ab2300cbecb23f3433bad888924c831ca',1,'BlockMergeSort::merge_partition()'],['../struct_kernel_multi_block_merge_sort.html#ab15895b4233aba0e279cc44a07a201fe',1,'KernelMultiBlockMergeSort::merge_partition()']]], + ['merge_5fstep_36',['merge_step',['../struct_block_merge_sort.html#ab65f190edf1851b37c39ad49ce99a43c',1,'BlockMergeSort']]], + ['meshgrid_37',['meshgrid',['../group__ops.html#ga577c911618575314de63d1060656a26e',1,'mlx::core']]], + ['metal_38',['metal',['../namespacemetal.html',1,'']]], + ['metal_2eh_39',['metal.h',['../metal_8h.html',1,'']]], + ['metal_3a_3afast_40',['fast',['../namespacemetal_1_1fast.html',1,'metal']]], + ['metal_3a_3aprecise_41',['precise',['../namespacemetal_1_1precise.html',1,'metal']]], + ['metal_5fimpl_2eh_42',['metal_impl.h',['../metal__impl_8h.html',1,'']]], + ['metal_5fkernel_43',['metal_kernel',['../namespacemlx_1_1core_1_1fast.html#ab16436b465dc10ce472193d541d8426e',1,'mlx::core::fast']]], + ['metalallocator_44',['MetalAllocator',['../classmlx_1_1core_1_1metal_1_1_metal_allocator.html',1,'mlx::core::metal']]], + ['metalkernelfunction_45',['MetalKernelFunction',['../namespacemlx_1_1core_1_1fast.html#a0e8c2c4ea7a946568c8fe5b4810417e0',1,'mlx::core::fast']]], + ['min_46',['Min',['../struct_min.html',1,'Min< U >'],['../classmlx_1_1core_1_1distributed_1_1_all_reduce.html#abb4560980e5d01aed14175ce8f6fc924a4f685dcd48e6614d6bb2ccda4f2686ef',1,'mlx::core::distributed::AllReduce::Min'],['../classmlx_1_1core_1_1_reduce.html#a0848518b16ae6d4043d6be247bdf31c9a0d3d1f5c94725bdc42fa692e2c074418',1,'mlx::core::Reduce::Min'],['../classmlx_1_1core_1_1_scan.html#a47bf2ec54ead4b8f00f9f188518630f1a7d2ee8f14f2e70a9d47170fecc6da898',1,'mlx::core::Scan::Min'],['../classmlx_1_1core_1_1_scatter.html#a614d19af11dc30644b2b4941033b613cad914e4c3475ce9858f2de4bf35dcfdbf',1,'mlx::core::Scatter::Min']]], + ['min_47',['min',['../struct_limits.html#a6e81584ba65a4dc6ff9366b458e3a20e',1,'Limits::min'],['../struct_limits_3_01uint8__t_01_4.html#a408bd5a337e7292f06e63da81193629a',1,'Limits< uint8_t >::min'],['../struct_limits_3_01uint16__t_01_4.html#ae173984c3be8b6750f27daed581805fe',1,'Limits< uint16_t >::min'],['../struct_limits_3_01uint32__t_01_4.html#ab0c3975e02053b234c7b606ababa66e1',1,'Limits< uint32_t >::min'],['../struct_limits_3_01uint64__t_01_4.html#a80627f39e951398283942cefa48f4dd0',1,'Limits< uint64_t >::min'],['../struct_limits_3_01int8__t_01_4.html#a7a809307d2bba80382f0645d277eaa4b',1,'Limits< int8_t >::min'],['../struct_limits_3_01int16__t_01_4.html#adca7139647801e223c35b0abc7da5240',1,'Limits< int16_t >::min'],['../struct_limits_3_01int32__t_01_4.html#af336a1b22a8ed6a83a4cfb5bf8869771',1,'Limits< int32_t >::min'],['../struct_limits_3_01int64__t_01_4.html#a1c90fb96af515badaccaa835b08f7428',1,'Limits< int64_t >::min'],['../struct_limits_3_01half_01_4.html#aca7b036c257878bf1b80912fb5d4516d',1,'Limits< half >::min'],['../struct_limits_3_01float_01_4.html#a3225e334d372ee86128c89a440d8648f',1,'Limits< float >::min'],['../struct_limits_3_01bfloat16__t_01_4.html#a2fd1811b9f615b2b897904bc27d1cb49',1,'Limits< bfloat16_t >::min'],['../struct_limits_3_01bool_01_4.html#a139f787b57536d455490b8ef801d37cc',1,'Limits< bool >::min'],['../struct_limits_3_01complex64__t_01_4.html#aa67b04aa7abcd67f7af0808737ab8e14',1,'Limits< complex64_t >::min'],['../structmetal_1_1__numeric__limits__impl_3_01bfloat16__t_01_4.html#adaed80031f5ca0ff69d30ec4c5d0c98f',1,'metal::_numeric_limits_impl< bfloat16_t >::min()'],['../namespacemetal.html#a6653b28c9473087141eddce39878d4d3',1,'metal::min()'],['../namespacemetal_1_1fast.html#a3e958e56a4712687c381a0b64d123e61',1,'metal::fast::min()'],['../namespacemetal_1_1precise.html#afed0da2f7df3505b5dffa2389c3cb36e',1,'metal::precise::min()'],['../group__ops.html#gab27599802617a4c8f9964ab5f4ffee12',1,'mlx::core::min(const array &a, bool keepdims, StreamOrDevice s={})'],['../group__ops.html#ga0140b91e9cdfc3fef0da8e332f65a9e8',1,'mlx::core::min(const array &a, StreamOrDevice s={})'],['../group__ops.html#ga6efb83cd46436678c8f8c4af15cc00f5',1,'mlx::core::min(const array &a, const std::vector< int > &axes, bool keepdims=false, StreamOrDevice s={})'],['../group__ops.html#ga36fa315eef677f4143868f552cd26d03',1,'mlx::core::min(const array &a, int axis, bool keepdims=false, StreamOrDevice s={})']]], + ['min3_48',['min3',['../namespacemetal.html#a005510c8c0f964ce2b8aad3ba76a7a3f',1,'metal::min3()'],['../namespacemetal_1_1fast.html#a606a4c1b34ce05ea89ca5af81724036f',1,'metal::fast::min3()'],['../namespacemetal_1_1precise.html#a4d37ce31c3549ca4772a4ee29798e231',1,'metal::precise::min3()']]], + ['min_5fexponent_49',['min_exponent',['../structmetal_1_1__numeric__limits__impl_3_01bfloat16__t_01_4.html#a13829f8c7a7c0efdc8946eff5d3c9470',1,'metal::_numeric_limits_impl< bfloat16_t >']]], + ['min_5fexponent10_50',['min_exponent10',['../structmetal_1_1__numeric__limits__impl_3_01bfloat16__t_01_4.html#aeaed172780720e06b8731cef3177e277',1,'metal::_numeric_limits_impl< bfloat16_t >']]], + ['minimum_51',['Minimum',['../struct_minimum.html',1,'Minimum'],['../structmlx_1_1core_1_1detail_1_1_minimum.html',1,'mlx::core::detail::Minimum'],['../classmlx_1_1core_1_1_minimum.html',1,'mlx::core::Minimum'],['../classmlx_1_1core_1_1_minimum.html#ab0f2ce17108df44b82cff68886b0f6f5',1,'mlx::core::Minimum::Minimum()']]], + ['minimum_52',['minimum',['../group__ops.html#ga49ba00c090f81f331c91b0c97040bce0',1,'mlx::core']]], + ['mlx_53',['mlx',['../namespacemlx.html',1,'']]], + ['mlx_2eh_54',['mlx.h',['../mlx_8h.html',1,'']]], + ['mlx_3a_3acore_55',['core',['../namespacemlx_1_1core.html',1,'mlx']]], + ['mlx_3a_3acore_3a_3aallocator_56',['allocator',['../namespacemlx_1_1core_1_1allocator.html',1,'mlx::core']]], + ['mlx_3a_3acore_3a_3adetail_57',['detail',['../namespacemlx_1_1core_1_1detail.html',1,'mlx::core']]], + ['mlx_3a_3acore_3a_3adistributed_58',['distributed',['../namespacemlx_1_1core_1_1distributed.html',1,'mlx::core']]], + ['mlx_3a_3acore_3a_3adistributed_3a_3adetail_59',['detail',['../namespacemlx_1_1core_1_1distributed_1_1detail.html',1,'mlx::core::distributed']]], + ['mlx_3a_3acore_3a_3afast_60',['fast',['../namespacemlx_1_1core_1_1fast.html',1,'mlx::core']]], + ['mlx_3a_3acore_3a_3afft_61',['fft',['../namespacemlx_1_1core_1_1fft.html',1,'mlx::core']]], + ['mlx_3a_3acore_3a_3aio_62',['io',['../namespacemlx_1_1core_1_1io.html',1,'mlx::core']]], + ['mlx_3a_3acore_3a_3alinalg_63',['linalg',['../namespacemlx_1_1core_1_1linalg.html',1,'mlx::core']]], + ['mlx_3a_3acore_3a_3ametal_64',['metal',['../namespacemlx_1_1core_1_1metal.html',1,'mlx::core']]], + ['mlx_3a_3acore_3a_3arandom_65',['random',['../namespacemlx_1_1core_1_1random.html',1,'mlx::core']]], + ['mlx_3a_3acore_3a_3ascheduler_66',['scheduler',['../namespacemlx_1_1core_1_1scheduler.html',1,'mlx::core']]], + ['mlx_3a_3asteel_67',['steel',['../namespacemlx_1_1steel.html',1,'mlx']]], + ['mlx_5fatomic_68',['mlx_atomic',['../structmlx__atomic.html',1,'']]], + ['mlx_5fatomic_3c_20t_2c_20enable_5fif_5ft_3c_20is_5fmetal_5fatomic_3c_20t_20_3e_20_3e_20_3e_69',['mlx_atomic< T, enable_if_t< is_metal_atomic< T > > >',['../structmlx__atomic_3_01_t_00_01enable__if__t_3_01is__metal__atomic_3_01_t_01_4_01_4_01_4.html',1,'']]], + ['mlx_5fatomic_5fcompare_5fexchange_5fweak_5fexplicit_70',['mlx_atomic_compare_exchange_weak_explicit',['../atomic_8h.html#ad7f32327ff66354cfa2f0cfdac79316f',1,'mlx_atomic_compare_exchange_weak_explicit(device mlx_atomic< T > *object, thread T *expected, T val, size_t offset): atomic.h'],['../atomic_8h.html#aa8f47b2e9b95d4b00ad51f08b070deb5',1,'mlx_atomic_compare_exchange_weak_explicit(device mlx_atomic< T > *object, thread uint *expected, uint val, size_t offset): atomic.h']]], + ['mlx_5fatomic_5ffetch_5fadd_5fexplicit_71',['mlx_atomic_fetch_add_explicit',['../atomic_8h.html#aad448d9e06e001700b65ca8317216a3b',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fand_5fexplicit_72',['mlx_atomic_fetch_and_explicit',['../atomic_8h.html#a253e3c870c0ddc7c28ab2f6ca2c3eae5',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fmax_5fexplicit_73',['mlx_atomic_fetch_max_explicit',['../atomic_8h.html#ac480f2b459a8ad9095cee353e152d00c',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fmax_5fexplicit_3c_20float_20_3e_74',['mlx_atomic_fetch_max_explicit< float >',['../atomic_8h.html#a1dce2abfa16417122c4d2bf261129ae4',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fmin_5fexplicit_75',['mlx_atomic_fetch_min_explicit',['../atomic_8h.html#a2ec33dca0039bd944d73d1c2b378cc19',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fmin_5fexplicit_3c_20float_20_3e_76',['mlx_atomic_fetch_min_explicit< float >',['../atomic_8h.html#ab7d1dc49f319f239b7ee0b7c72976dd0',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fmul_5fexplicit_77',['mlx_atomic_fetch_mul_explicit',['../atomic_8h.html#adfdbea60436f14f1af9ce36e2a0a77a3',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5for_5fexplicit_78',['mlx_atomic_fetch_or_explicit',['../atomic_8h.html#ab7391f197001471e4788312bdb6ab37a',1,'atomic.h']]], + ['mlx_5fatomic_5fload_5fexplicit_79',['mlx_atomic_load_explicit',['../atomic_8h.html#a253a4e8c2c5768a069e2791b627dfc99',1,'atomic.h']]], + ['mlx_5fatomic_5fstore_5fexplicit_80',['mlx_atomic_store_explicit',['../atomic_8h.html#a0ae453140b0819a4c02f265334de98c0',1,'atomic.h']]], + ['mlx_5flapack_5ffunc_81',['MLX_LAPACK_FUNC',['../lapack_8h.html#ae22db9704827bf013a0a61f21a47464b',1,'lapack.h']]], + ['mlx_5fmtl_5fconst_82',['MLX_MTL_CONST',['../kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd',1,'MLX_MTL_CONST: gemv_masked.h'],['../quantized_8h.html#a0386011c52d03e60885a31e6fbd903dd',1,'MLX_MTL_CONST: quantized.h'],['../sort_8h.html#a0386011c52d03e60885a31e6fbd903dd',1,'MLX_MTL_CONST: sort.h']]], + ['mlx_5fmtl_5floop_5funroll_83',['MLX_MTL_LOOP_UNROLL',['../sort_8h.html#ad34b622323cebef136669fedd7229515',1,'sort.h']]], + ['mlx_5fmtl_5fpragma_5funroll_84',['MLX_MTL_PRAGMA_UNROLL',['../kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3',1,'MLX_MTL_PRAGMA_UNROLL: gemv_masked.h'],['../backend_2metal_2kernels_2utils_8h.html#a069b682d7d21827461544817d722bfd3',1,'MLX_MTL_PRAGMA_UNROLL: utils.h']]], + ['mlxconvparams_85',['MLXConvParams',['../struct_m_l_x_conv_params.html',1,'']]], + ['mlxconvparams_3c_202_20_3e_86',['MLXConvParams< 2 >',['../struct_m_l_x_conv_params.html',1,'']]], + ['mlxfastattentionparams_87',['MLXFastAttentionParams',['../struct_m_l_x_fast_attention_params.html',1,'']]], + ['mlxscaleddotproductattentionparams_88',['MLXScaledDotProductAttentionParams',['../struct_m_l_x_scaled_dot_product_attention_params.html',1,'']]], + ['mma_89',['mma',['../structmlx_1_1steel_1_1_base_m_m_a_frag_3_01_t_00_018_00_018_01_4.html#a8028512f5a3d2b6acaf966be529627a3',1,'mlx::steel::BaseMMAFrag< T, 8, 8 >::mma(thread frag_type &D, thread frag_type &A, thread frag_type &B, thread frag_type &C)'],['../structmlx_1_1steel_1_1_base_m_m_a_frag_3_01_t_00_018_00_018_01_4.html#a1868f57d57c8adedab2c58492ec76946',1,'mlx::steel::BaseMMAFrag< T, 8, 8 >::mma(thread mat_type &D, thread mat_type &A, thread mat_type &B, thread mat_type &C)'],['../structmlx_1_1steel_1_1_block_m_m_a.html#a6a2c2a6d5e767d52c41b42a9d36086b0',1,'mlx::steel::BlockMMA::mma()']]], + ['mma_2eh_90',['mma.h',['../mma_8h.html',1,'']]], + ['mma_5ft_91',['mma_t',['../structmlx_1_1steel_1_1_g_e_m_m_kernel.html#add8c6a31011a4895667c2a94a5af3782',1,'mlx::steel::GEMMKernel']]], + ['mmafrag_5facc_5ft_92',['MMAFrag_acc_t',['../structmlx_1_1steel_1_1_block_m_m_a.html#ae2c42cb6d0dde785859164c195f4d13c',1,'mlx::steel::BlockMMA']]], + ['mmafrag_5ft_93',['MMAFrag_t',['../structmlx_1_1steel_1_1_m_m_a_tile.html#abe33de70e34300745bad9aa822fd0382',1,'mlx::steel::MMATile']]], + ['mmatile_94',['MMATile',['../structmlx_1_1steel_1_1_m_m_a_tile.html',1,'mlx::steel::MMATile< T, kTileRows_, kTileCols_, MMAFrag_ >'],['../structmlx_1_1steel_1_1_m_m_a_tile.html#aa3fb310dd08ec23c334511f7b316d1b6',1,'mlx::steel::MMATile::MMATile()']]], + ['mmatile_3c_20float_2c_201_2c_20tn_2c_20mlx_3a_3asteel_3a_3abasemmafrag_20_3e_95',['MMATile< float, 1, TN, mlx::steel::BaseMMAFrag >',['../structmlx_1_1steel_1_1_m_m_a_tile.html',1,'mlx::steel']]], + ['mmatile_3c_20float_2c_20tm_2c_201_2c_20mlx_3a_3asteel_3a_3abasemmafrag_20_3e_96',['MMATile< float, TM, 1, mlx::steel::BaseMMAFrag >',['../structmlx_1_1steel_1_1_m_m_a_tile.html',1,'mlx::steel']]], + ['mmatile_3c_20float_2c_20tm_2c_20tn_2c_20mlx_3a_3asteel_3a_3abasemmafrag_20_3e_97',['MMATile< float, TM, TN, mlx::steel::BaseMMAFrag >',['../structmlx_1_1steel_1_1_m_m_a_tile.html',1,'mlx::steel']]], + ['move_5fshared_5fbuffer_98',['move_shared_buffer',['../classmlx_1_1core_1_1array.html#acce00db63e0f3d80f797b02397ade836',1,'mlx::core::array::move_shared_buffer(array other, const std::vector< size_t > &strides, Flags flags, size_t data_size, size_t offset=0)'],['../classmlx_1_1core_1_1array.html#a38d7ad605f8282e5e49d0c09e0555c78',1,'mlx::core::array::move_shared_buffer(array other)']]], + ['moveaxis_99',['moveaxis',['../group__ops.html#ga24067d10a842db2c9d509ea48135a2c3',1,'mlx::core']]], + ['mpinplace_100',['MPINPLACE',['../namespacepocketfft_1_1detail.html#af5eedf3cdfc83c0a30807092c39a9ce2',1,'pocketfft::detail']]], + ['mtl_5fconst_101',['MTL_CONST',['../defines_8h.html#a767ed9f2604de22b259cee02c4ce1d22',1,'defines.h']]], + ['mtl_5fdevice_102',['mtl_device',['../classmlx_1_1core_1_1metal_1_1_device.html#a31dba377f2be44a746db10d1b9367653',1,'mlx::core::metal::Device']]], + ['mtl_5fresidency_5fset_103',['mtl_residency_set',['../classmlx_1_1core_1_1metal_1_1_residency_set.html#ac4bfe5ef5e2eaebc458a1ed1953d15e9',1,'mlx::core::metal::ResidencySet']]], + ['mtlfclist_104',['MTLFCList',['../namespacemlx_1_1core_1_1metal.html#a616e09a1ef321d527770721cef264c54',1,'mlx::core::metal']]], + ['mtx_105',['mtx',['../structmlx_1_1core_1_1scheduler_1_1_stream_thread.html#a70410c9e612f871663929f1e8441a976',1,'mlx::core::scheduler::StreamThread']]], + ['multi_5fiter_106',['multi_iter',['../classpocketfft_1_1detail_1_1multi__iter.html',1,'pocketfft::detail::multi_iter< N >'],['../classpocketfft_1_1detail_1_1multi__iter.html#a9be43bb18840202da6d17988fccc64b9',1,'pocketfft::detail::multi_iter::multi_iter()']]], + ['multiply_107',['Multiply',['../structmlx_1_1core_1_1detail_1_1_multiply.html',1,'mlx::core::detail::Multiply'],['../classmlx_1_1core_1_1_multiply.html',1,'mlx::core::Multiply'],['../struct_multiply.html',1,'Multiply'],['../classmlx_1_1core_1_1_multiply.html#aca5c50f900321f3eb4d6fbcbc225c00c',1,'mlx::core::Multiply::Multiply()']]], + ['multiply_108',['multiply',['../group__ops.html#gaf57392e641640b5d06e4c99518391c38',1,'mlx::core']]], + ['multivariate_5fnormal_109',['multivariate_normal',['../namespacemlx_1_1core_1_1random.html#a8c37da3c1c0c561cad7499d6d9db81fb',1,'mlx::core::random']]] ]; diff --git a/docs/build/html/search/functions_1.js b/docs/build/html/search/functions_1.js index 255ad1343..b4193ba2a 100644 --- a/docs/build/html/search/functions_1.js +++ b/docs/build/html/search/functions_1.js @@ -22,7 +22,7 @@ var searchData= ['all_19',['all',['../group__ops.html#ga3b1b90ef1275ca17655b6d7f25d3ee68',1,'mlx::core::all(const array &a, bool keepdims, StreamOrDevice s={})'],['../group__ops.html#ga3689e12e8f42dadb4cbe2b07dc4099f4',1,'mlx::core::all(const array &a, StreamOrDevice s={})'],['../group__ops.html#gac0919c6ba53aea35a7683dea7e9a9a59',1,'mlx::core::all(const array &a, const std::vector< int > &axes, bool keepdims=false, StreamOrDevice s={})'],['../group__ops.html#gae2d5fcc5b62d673cca76c08b7b4afbbc',1,'mlx::core::all(const array &a, int axis, bool keepdims=false, StreamOrDevice s={})']]], ['all_5fgather_20',['all_gather',['../namespacemlx_1_1core_1_1distributed_1_1detail.html#aeb5a1726358213bc75756506f7b54d04',1,'mlx::core::distributed::detail::all_gather()'],['../namespacemlx_1_1core_1_1distributed.html#a82ef5e8cc7ac62cd228e51b1c1b77cb7',1,'mlx::core::distributed::all_gather()']]], ['all_5freduce_21',['all_reduce',['../reduce__all_8h.html#a99ef48ae72b3e715c5f4d7ea07cd213d',1,'reduce_all.h']]], - ['all_5freduce_5fdispatch_22',['all_reduce_dispatch',['../namespacemlx_1_1core.html#af7b7ca7c6aa87558d9f98cee5c7a99a8',1,'mlx::core']]], + ['all_5freduce_5fdispatch_22',['all_reduce_dispatch',['../namespacemlx_1_1core.html#a3ab0fd997d9a35782106ff083a72e098',1,'mlx::core']]], ['all_5fsum_23',['all_sum',['../namespacemlx_1_1core_1_1distributed_1_1detail.html#aa1d225b25f7b6426c48c5e35860ee960',1,'mlx::core::distributed::detail::all_sum()'],['../namespacemlx_1_1core_1_1distributed.html#a67ccb1a5445fc6f5db49dd36a15e5980',1,'mlx::core::distributed::all_sum()']]], ['allclose_24',['allclose',['../group__ops.html#gaf0cd4257de7542daf9faf5e605e31020',1,'mlx::core']]], ['allgather_25',['AllGather',['../classmlx_1_1core_1_1distributed_1_1_all_gather.html#af4b10a5b61f160fb64353057c185b661',1,'mlx::core::distributed::AllGather']]], diff --git a/docs/build/html/search/functions_11.js b/docs/build/html/search/functions_11.js index 7dfd15ac3..f07193647 100644 --- a/docs/build/html/search/functions_11.js +++ b/docs/build/html/search/functions_11.js @@ -22,5 +22,6 @@ var searchData= ['quantizedmatmul_19',['QuantizedMatmul',['../classmlx_1_1core_1_1_quantized_matmul.html#a5bd164d038d9dc21919f7e0bfdeaa25c',1,'mlx::core::QuantizedMatmul']]], ['quiet_5fnan_20',['quiet_NaN',['../structmetal_1_1__numeric__limits__impl_3_01bfloat16__t_01_4.html#aebeb07c01984be246bc2d1b8f8e4ac7b',1,'metal::_numeric_limits_impl< bfloat16_t >']]], ['qvm_21',['qvm',['../quantized_8h.html#ad84f7d5ab9e32dbbe3ca759ae5d5d5c5',1,'quantized.h']]], - ['qvm_5fimpl_22',['qvm_impl',['../quantized_8h.html#a4a8c8db7d5d480733726fd6d1a645e12',1,'quantized.h']]] + ['qvm_5fimpl_22',['qvm_impl',['../quantized_8h.html#a1546533c5b925b2fbb3bec870ec7487a',1,'quantized.h']]], + ['qvm_5fsplit_5fk_23',['qvm_split_k',['../quantized_8h.html#ab8243818512d6078d23e6ffb65fd7bb8',1,'quantized.h']]] ]; diff --git a/docs/build/html/search/functions_13.js b/docs/build/html/search/functions_13.js index bbac045cb..5342f6271 100644 --- a/docs/build/html/search/functions_13.js +++ b/docs/build/html/search/functions_13.js @@ -17,7 +17,7 @@ var searchData= ['scatter_5fprod_14',['scatter_prod',['../group__ops.html#ga3708b5bcb61e2c63d213c4ce6ad0ffc0',1,'mlx::core::scatter_prod(const array &a, const std::vector< array > &indices, const array &updates, const std::vector< int > &axes, StreamOrDevice s={})'],['../group__ops.html#gaf83c53c453faa9083ba27e4b97539339',1,'mlx::core::scatter_prod(const array &a, const array &indices, const array &updates, int axis, StreamOrDevice s={})']]], ['scheduler_15',['Scheduler',['../classmlx_1_1core_1_1scheduler_1_1_scheduler.html#a3ae42aed78a2200e9d02776fcd2316ba',1,'mlx::core::scheduler::Scheduler::Scheduler()'],['../classmlx_1_1core_1_1scheduler_1_1_scheduler.html#a61a74e3628899e66dde600e24a750648',1,'mlx::core::scheduler::Scheduler::Scheduler(const Scheduler &)=delete'],['../classmlx_1_1core_1_1scheduler_1_1_scheduler.html#ac3f77b7c93220dadd0b3bb2e903b7059',1,'mlx::core::scheduler::Scheduler::Scheduler(Scheduler &&)=delete']]], ['scheduler_16',['scheduler',['../namespacemlx_1_1core_1_1scheduler.html#ae856e468c2f7c8f8ec672522cc13730b',1,'mlx::core::scheduler']]], - ['sdpa_5fvector_17',['sdpa_vector',['../sdpa__vector_8h.html#a6f0d7918430064bab910bdaa6c64e927',1,'sdpa_vector.h']]], + ['sdpa_5fvector_17',['sdpa_vector',['../sdpa__vector_8h.html#a4bf36f16e16c1c62d9b243573568e5ae',1,'sdpa_vector.h']]], ['seed_18',['seed',['../classmlx_1_1core_1_1random_1_1_key_sequence.html#a9f19c5da2031cba50d0ff996924347d8',1,'mlx::core::random::KeySequence::seed()'],['../namespacemlx_1_1core_1_1random.html#ac4ad325b613257306df74595d3d0e23b',1,'mlx::core::random::seed()']]], ['seek_19',['seek',['../structmlx_1_1core_1_1_contiguous_iterator.html#a24719ee9e8667885d29c2ad74445520c',1,'mlx::core::ContiguousIterator::seek()'],['../classmlx_1_1core_1_1io_1_1_reader.html#acea55078bd39ccaa27a9a36f17a39cd1',1,'mlx::core::io::Reader::seek()'],['../classmlx_1_1core_1_1io_1_1_writer.html#a9c1716dda53aa36faea9c8fb1a3e34d4',1,'mlx::core::io::Writer::seek()'],['../classmlx_1_1core_1_1io_1_1_parallel_file_reader.html#a673c16b669f3cee13f387b7b0a1f39f7',1,'mlx::core::io::ParallelFileReader::seek()'],['../classmlx_1_1core_1_1io_1_1_file_writer.html#a9646f4ea048ae58719daeb588e2de433',1,'mlx::core::io::FileWriter::seek()']]], ['select_20',['Select',['../classmlx_1_1core_1_1_select.html#a6f833fe55dd68ad3726bbf9a8f75eec9',1,'mlx::core::Select']]], diff --git a/docs/build/html/search/functions_3.js b/docs/build/html/search/functions_3.js index 29bcd7684..2e834b06f 100644 --- a/docs/build/html/search/functions_3.js +++ b/docs/build/html/search/functions_3.js @@ -18,85 +18,87 @@ var searchData= ['clip_15',['clip',['../group__ops.html#ga157cd7c23f9b306fee2e1eb2b9bf1dd8',1,'mlx::core']]], ['cmplx_16',['cmplx',['../structpocketfft_1_1detail_1_1cmplx.html#a5b1ce506f1023f5254025ac81b831a2c',1,'pocketfft::detail::cmplx::cmplx()'],['../structpocketfft_1_1detail_1_1cmplx.html#a05491b4f1f22ca0bc49012f6a1c1710a',1,'pocketfft::detail::cmplx::cmplx(T r_, T i_)']]], ['cndarr_17',['cndarr',['../classpocketfft_1_1detail_1_1cndarr.html#abf73f1b4ddcfb27d7f85cfa441607129',1,'pocketfft::detail::cndarr']]], - ['col_5freduce_5flooped_18',['col_reduce_looped',['../reduce__col_8h.html#a11bfc6112ae2386ac03f5ea7b7d93385',1,'reduce_col.h']]], - ['col_5freduce_5fsmall_19',['col_reduce_small',['../reduce__col_8h.html#adf7aeb18cd1d5042cf6d9b46b582d8ce',1,'reduce_col.h']]], - ['collapse_5fcontiguous_5fdims_20',['collapse_contiguous_dims',['../namespacemlx_1_1core.html#a38fe6ec5220d13d96c7dad7556d2b613',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< std::vector< int64_t > > &strides, int64_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#af2895f9b0083efd8221275eb8cadccbe',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< std::vector< size_t > > &strides, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a90e2b6edc0fe82230cb93f5ea39febb4',1,'mlx::core::collapse_contiguous_dims(const std::vector< array > &xs, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#ac813412cce77fc1340dcfefc6e099276',1,'mlx::core::collapse_contiguous_dims(Arrays &&... xs)'],['../namespacemlx_1_1core.html#aab3cc7f3808934ae0727b920eba231bd',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< int64_t > &strides, int64_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a1e0cbcf109d32794ffc8efc7302ba9b0',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< size_t > &strides, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a4ee50bfb240512d0c0ce151dfe2c74ef',1,'mlx::core::collapse_contiguous_dims(const array &a, size_t size_cap=std::numeric_limits< int32_t >::max())']]], - ['commandencoder_21',['CommandEncoder',['../structmlx_1_1core_1_1metal_1_1_command_encoder.html#a2334774486f447213ee997e55c2e52a3',1,'mlx::core::metal::CommandEncoder::CommandEncoder(MTL::CommandBuffer *cbuf)'],['../structmlx_1_1core_1_1metal_1_1_command_encoder.html#ac68ca977b5bde5434284ce7979647f14',1,'mlx::core::metal::CommandEncoder::CommandEncoder(const CommandEncoder &)=delete']]], - ['commit_5fcommand_5fbuffer_22',['commit_command_buffer',['../classmlx_1_1core_1_1metal_1_1_device.html#a95248f1387824067fd4fed23ace5ac0c',1,'mlx::core::metal::Device']]], - ['communication_5fstream_23',['communication_stream',['../namespacemlx_1_1core_1_1distributed_1_1detail.html#ac3612edf0e0e18c1e4ba0ce7c6e35cd6',1,'mlx::core::distributed::detail']]], - ['compile_24',['compile',['../namespacemlx_1_1core.html#a3ac798e65e59fe10b7fb5c522efce782',1,'mlx::core::compile()'],['../namespacemlx_1_1core_1_1detail.html#ac3b7b09892ff7290d5f3ef26cb444329',1,'mlx::core::detail::compile(const std::function< std::vector< array >(const std::vector< array > &)> &fun, std::uintptr_t fun_id, bool shapeless=false, std::vector< uint64_t > constants={})']]], - ['compile_5favailable_5ffor_5fdevice_25',['compile_available_for_device',['../namespacemlx_1_1core_1_1detail.html#aeeff2ba6ec3d9d4ed090de6d2681dbc2',1,'mlx::core::detail']]], - ['compile_5fclear_5fcache_26',['compile_clear_cache',['../namespacemlx_1_1core_1_1detail.html#a3fb927c209b946aefebb195993fbe4cf',1,'mlx::core::detail']]], - ['compile_5ferase_27',['compile_erase',['../namespacemlx_1_1core_1_1detail.html#a69eb76a14f845ca000f1ccb2edda0175',1,'mlx::core::detail']]], - ['compiled_28',['Compiled',['../classmlx_1_1core_1_1_compiled.html#a2d8cefff835c419a48a077d306b8e051',1,'mlx::core::Compiled']]], - ['compiled_5fallocate_5foutputs_29',['compiled_allocate_outputs',['../namespacemlx_1_1core.html#ab8c3c4fc05745f586de922c8266f4fce',1,'mlx::core']]], - ['compiled_5fcheck_5fcontiguity_30',['compiled_check_contiguity',['../namespacemlx_1_1core.html#a3b900ab319948c5a01a3ecd30a709027',1,'mlx::core']]], - ['complex128_5ft_31',['complex128_t',['../structmlx_1_1core_1_1complex128__t.html#aa15d0b805f8790f7c7b76fc7b9d677e0',1,'mlx::core::complex128_t::complex128_t(double v, double u)'],['../structmlx_1_1core_1_1complex128__t.html#abf2842253b874f9f13f39ea68a89e5b6',1,'mlx::core::complex128_t::complex128_t(std::complex< double > v)'],['../structmlx_1_1core_1_1complex128__t.html#a526fba96d7e815360cb4226af085a1bf',1,'mlx::core::complex128_t::complex128_t(T x)']]], - ['complex64_5ft_32',['complex64_t',['../structcomplex64__t.html#adbd392a5e92d31997380ad0a38be4be8',1,'complex64_t::complex64_t(float real, float imag)'],['../structcomplex64__t.html#a29782289bb90d6294099667b86509cd3',1,'complex64_t::complex64_t()'],['../structcomplex64__t.html#a905b048d70eb8d748a62454268242291',1,'complex64_t::complex64_t() threadgroup'],['../structcomplex64__t.html#a33a2452eb33b5ed53655773539c357a5',1,'complex64_t::complex64_t(T x) thread'],['../structcomplex64__t.html#a89b65ace8588b7bf215355f705eb23d9',1,'complex64_t::complex64_t(T x) threadgroup'],['../structcomplex64__t.html#ac81b486f642fb3b26c5d659917bdbcd0',1,'complex64_t::complex64_t(T x) device'],['../structcomplex64__t.html#a0a27a41206400f1e62b60ceb56960c93',1,'complex64_t::complex64_t(T x) const ant'],['../structmlx_1_1core_1_1complex64__t.html#a697cc973ae27d63c8e00d830e780bd8c',1,'mlx::core::complex64_t::complex64_t(float v, float u)'],['../structmlx_1_1core_1_1complex64__t.html#ae065e39938f9c4374b4116f4c67d4d09',1,'mlx::core::complex64_t::complex64_t(std::complex< float > v)'],['../structmlx_1_1core_1_1complex64__t.html#a2232cbbe591a9d2bc228cb23fac38b50',1,'mlx::core::complex64_t::complex64_t(T x)']]], - ['complex_5fmul_33',['complex_mul',['../radix_8h.html#a5bfc53b531214c9ce277bebc18aa67d6',1,'radix.h']]], - ['complex_5fmul_5fconj_34',['complex_mul_conj',['../radix_8h.html#a0e2dfd3d1dda09f47ccc64eec35629f3',1,'radix.h']]], - ['compute_5fstrided_5findices_35',['compute_strided_indices',['../struct_read_writer.html#a7c903fbb8b85a856ba5564d7df537cdf',1,'ReadWriter']]], - ['concatenate_36',['Concatenate',['../classmlx_1_1core_1_1_concatenate.html#acff07853de2d31faeec7c4ca40ce0888',1,'mlx::core::Concatenate']]], - ['concatenate_37',['concatenate',['../group__ops.html#gabdc36fa65697d0361c8d67495de77129',1,'mlx::core::concatenate(const std::vector< array > &arrays, int axis, StreamOrDevice s={})'],['../group__ops.html#gaa95c34ca3a8877f2c50cb60e7fa312b8',1,'mlx::core::concatenate(const std::vector< array > &arrays, StreamOrDevice s={})']]], - ['concatenate_5fgpu_38',['concatenate_gpu',['../namespacemlx_1_1core.html#a050299d0d366ca5c9d09d1004dcc3e7d',1,'mlx::core']]], - ['concurrentcontext_39',['ConcurrentContext',['../structmlx_1_1core_1_1metal_1_1_command_encoder_1_1_concurrent_context.html#aee044d7729739c96e845823f9ecc5174',1,'mlx::core::metal::CommandEncoder::ConcurrentContext']]], - ['conj_40',['conj',['../namespacepocketfft_1_1detail.html#a66d79051d502046a9b9f103e744dbad3',1,'pocketfft::detail']]], - ['conjugate_41',['Conjugate',['../classmlx_1_1core_1_1_conjugate.html#a627f9e6a8729fb3ffb3ca3228d007c87',1,'mlx::core::Conjugate']]], - ['conjugate_42',['conjugate',['../group__ops.html#ga5b596906bf8cdc8d97ed6ddc9aeb4c23',1,'mlx::core']]], - ['contiguous_5fscan_43',['contiguous_scan',['../scan_8h.html#a60d279b9add7d56639bb209408f09d79',1,'scan.h']]], - ['contiguousiterator_44',['ContiguousIterator',['../structmlx_1_1core_1_1_contiguous_iterator.html#a68794af4a442d3d8ac4647817af8e1f6',1,'mlx::core::ContiguousIterator::ContiguousIterator()'],['../structmlx_1_1core_1_1_contiguous_iterator.html#a6cb378408b6f546eeb6ade1a4faafe3c',1,'mlx::core::ContiguousIterator::ContiguousIterator(const array &a)'],['../structmlx_1_1core_1_1_contiguous_iterator.html#a16bdacb53f65b7284068cd49d4cba292',1,'mlx::core::ContiguousIterator::ContiguousIterator(const std::vector< int > &shape, const std::vector< StrideT > &strides, int dims)']]], - ['conv_45',['conv',['../namespacemlx_1_1core_1_1metal.html#ab1704e853394c725668c06752ebb5c24',1,'mlx::core::metal']]], - ['conv1d_46',['conv1d',['../group__ops.html#ga30d47e08093c03a3676f235f9f559411',1,'mlx::core']]], - ['conv2d_47',['conv2d',['../group__ops.html#ga73b02833229678786e7f302d458d5a83',1,'mlx::core']]], - ['conv2dinputblockloadergeneral_48',['Conv2DInputBlockLoaderGeneral',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_general.html#a1d83af561a483432bf8dcb42e734b23b',1,'mlx::steel::Conv2DInputBlockLoaderGeneral']]], - ['conv2dinputblockloaderlargefilter_49',['Conv2DInputBlockLoaderLargeFilter',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_large_filter.html#a8755116a535539744e4947bc69f9c50f',1,'mlx::steel::Conv2DInputBlockLoaderLargeFilter']]], - ['conv2dinputblockloadersmallchannels_50',['Conv2DInputBlockLoaderSmallChannels',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_channels.html#ab9fd3fdeab94470dde3326f1dd5c455a',1,'mlx::steel::Conv2DInputBlockLoaderSmallChannels']]], - ['conv2dinputblockloadersmallfilter_51',['Conv2DInputBlockLoaderSmallFilter',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_filter.html#a0a2cbf57c51cd928722e3f06aafcf933',1,'mlx::steel::Conv2DInputBlockLoaderSmallFilter']]], - ['conv2dweightblockloader_52',['Conv2DWeightBlockLoader',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader.html#a9a7dca3512b64cffb6eac305d795831c',1,'mlx::steel::Conv2DWeightBlockLoader']]], - ['conv2dweightblockloadergeneral_53',['Conv2DWeightBlockLoaderGeneral',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_general.html#ad0550fabbdc9297559381a5b488e9af1',1,'mlx::steel::Conv2DWeightBlockLoaderGeneral']]], - ['conv2dweightblockloadersmallchannels_54',['Conv2DWeightBlockLoaderSmallChannels',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_small_channels.html#ae1806ea1c19713819dee83a38ab35fa6',1,'mlx::steel::Conv2DWeightBlockLoaderSmallChannels']]], - ['conv3d_55',['conv3d',['../group__ops.html#ga6e9907d2f14dc4803e4306b3dbc4b3ca',1,'mlx::core']]], - ['conv_5fgeneral_56',['conv_general',['../group__ops.html#ga2236e5dfc7e52e28abf6c21675d0a51e',1,'mlx::core::conv_general(array input, array weight, std::vector< int > stride={}, std::vector< int > padding_lo={}, std::vector< int > padding_hi={}, std::vector< int > kernel_dilation={}, std::vector< int > input_dilation={}, int groups=1, bool flip=false, StreamOrDevice s={})'],['../group__ops.html#gab59f89942cd1efaadffe9e8762e3c99d',1,'mlx::core::conv_general(const array &input, const array &weight, std::vector< int > stride={}, std::vector< int > padding={}, std::vector< int > kernel_dilation={}, std::vector< int > input_dilation={}, int groups=1, bool flip=false, StreamOrDevice s={})']]], - ['conv_5ftranspose1d_57',['conv_transpose1d',['../group__ops.html#gaa30bf1adcd78d1c2595d07b215731714',1,'mlx::core']]], - ['conv_5ftranspose2d_58',['conv_transpose2d',['../group__ops.html#gaebb59971cb9bc45005dc1d398e4f0a3d',1,'mlx::core']]], - ['conv_5ftranspose3d_59',['conv_transpose3d',['../group__ops.html#ga8db814da631d9cd32a8d6563bf4ac530',1,'mlx::core']]], - ['convolution_60',['Convolution',['../classmlx_1_1core_1_1_convolution.html#a6f1de77b719bb13217b0d8c64cabb8ef',1,'mlx::core::Convolution']]], - ['copy_61',['Copy',['../classmlx_1_1core_1_1_copy.html#a6243e044af119105ffaaed7d405cd584',1,'mlx::core::Copy']]], - ['copy_62',['copy',['../namespacemlx_1_1core.html#a479648542a2bea151b947b18f0e79dd2',1,'mlx::core::copy()'],['../namespacemlx_1_1core_1_1metal.html#aa215e631e2680f04a591b88d91571719',1,'mlx::core::metal::copy()'],['../group__ops.html#gae306e93af12f774bd80bad6c231b09d6',1,'mlx::core::copy()']]], - ['copy_5fg_63',['copy_g',['../metal_2kernels_2copy_8h.html#a778ce2dbfbaa23b24bd5efbe68448c36',1,'copy.h']]], - ['copy_5fg_5fnd1_64',['copy_g_nd1',['../metal_2kernels_2copy_8h.html#aba4530a7db6a61ca36f50e4f5e58fb77',1,'copy.h']]], - ['copy_5fg_5fnd2_65',['copy_g_nd2',['../metal_2kernels_2copy_8h.html#aee678c7c31119f3e609685589f37490c',1,'copy.h']]], - ['copy_5fg_5fnd3_66',['copy_g_nd3',['../metal_2kernels_2copy_8h.html#a821f8f3f3891159a295c66fc25aed1ff',1,'copy.h']]], - ['copy_5fgg_67',['copy_gg',['../metal_2kernels_2copy_8h.html#a1e39c2683eeaf05955e7619fbd34aea5',1,'copy.h']]], - ['copy_5fgg_5fnd1_68',['copy_gg_nd1',['../metal_2kernels_2copy_8h.html#a3278d9c999718bee3ccbe2922f501bf1',1,'copy.h']]], - ['copy_5fgg_5fnd2_69',['copy_gg_nd2',['../metal_2kernels_2copy_8h.html#a3e2d3cc7f34f56170409b6735f51a950',1,'copy.h']]], - ['copy_5fgg_5fnd3_70',['copy_gg_nd3',['../metal_2kernels_2copy_8h.html#a59f43b5bffed936d7559ceb06a10aabd',1,'copy.h']]], - ['copy_5fgpu_71',['copy_gpu',['../namespacemlx_1_1core.html#addaa46a13ac2deb1d9ce621338320e0e',1,'mlx::core::copy_gpu(const array &src, array &out, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#a6a6f4e46c8fc44fdc74c50ace02bcf38',1,'mlx::core::copy_gpu(const array &src, array &out, CopyType ctype)']]], - ['copy_5fgpu_5finplace_72',['copy_gpu_inplace',['../namespacemlx_1_1core.html#a69e30f5d30a6d72ac0ffe4886f24b7ba',1,'mlx::core::copy_gpu_inplace(const array &in, array &out, const std::vector< int > &data_shape, const std::vector< stride_t > &i_strides, const std::vector< stride_t > &o_strides, int64_t i_offset, int64_t o_offset, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#a8e1ccb0ed9387b0a789311d9f8964803',1,'mlx::core::copy_gpu_inplace(const array &src, array &out, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#ae55b801b09ccf55cba96278163a9b1ef',1,'mlx::core::copy_gpu_inplace(const array &in, array &out, const std::vector< int64_t > &istride, int64_t ioffset, CopyType ctype, const Stream &s)']]], - ['copy_5fhartley_73',['copy_hartley',['../namespacepocketfft_1_1detail.html#abac3fcc8ce83800d228774f64c28d4c3',1,'pocketfft::detail::copy_hartley(const multi_iter< vlen > &it, const vtype_t< T > *src, ndarr< T > &dst)'],['../namespacepocketfft_1_1detail.html#ae7b44d2773d9d06a9787aff01d66b3ed',1,'pocketfft::detail::copy_hartley(const multi_iter< vlen > &it, const T *src, ndarr< T > &dst)']]], - ['copy_5finplace_74',['copy_inplace',['../namespacemlx_1_1core.html#a98495894a796b2cc6d022e7a03432c64',1,'mlx::core::copy_inplace(const array &src, array &dst, CopyType ctype)'],['../namespacemlx_1_1core.html#aad636e2d0b2f882cadd1b438f4daa9ed',1,'mlx::core::copy_inplace(const array &src, array &dst, const std::vector< int > &data_shape, const std::vector< stride_t > &i_strides, const std::vector< stride_t > &o_strides, int64_t i_offset, int64_t o_offset, CopyType ctype)']]], - ['copy_5finput_75',['copy_input',['../namespacepocketfft_1_1detail.html#aff05be3064743c1143b19318ab12ad4a',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< cmplx< T > > &src, cmplx< vtype_t< T > > *dst)'],['../namespacepocketfft_1_1detail.html#a30fc708f9d8f9cfa74194925c7863c0a',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< T > &src, vtype_t< T > *dst)'],['../namespacepocketfft_1_1detail.html#a3387bd35f237870e42b8461769e6aec4',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< T > &src, T *dst)']]], - ['copy_5foutput_76',['copy_output',['../namespacepocketfft_1_1detail.html#a1523a037300a8da05db210b802d9cb0e',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const cmplx< vtype_t< T > > *src, ndarr< cmplx< T > > &dst)'],['../namespacepocketfft_1_1detail.html#a21980853aca4d92ed06e3dcffe7ef660',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const vtype_t< T > *src, ndarr< T > &dst)'],['../namespacepocketfft_1_1detail.html#a310481c334e46674710ba794ad7403c0',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const T *src, ndarr< T > &dst)']]], - ['copy_5fs_77',['copy_s',['../metal_2kernels_2copy_8h.html#aef09f9b9475345b1bba121d037d222ea',1,'copy.h']]], - ['copy_5fs2_78',['copy_s2',['../metal_2kernels_2copy_8h.html#a8023e9335cc5334847a8d315042be3a3',1,'copy.h']]], - ['copy_5fshared_5fbuffer_79',['copy_shared_buffer',['../classmlx_1_1core_1_1array.html#a28df7a333d90a311c49bc4bce7a1ad6d',1,'mlx::core::array::copy_shared_buffer(const array &other, const std::vector< size_t > &strides, Flags flags, size_t data_size, size_t offset=0)'],['../classmlx_1_1core_1_1array.html#a92974c656c35a972ad241f80584bbd29',1,'mlx::core::array::copy_shared_buffer(const array &other)']]], - ['copy_5fv_80',['copy_v',['../metal_2kernels_2copy_8h.html#ae26a13e0c8e6c15f7b10078e65970659',1,'copy.h']]], - ['copy_5fv2_81',['copy_v2',['../metal_2kernels_2copy_8h.html#aee14a5326f53d9b30b0b38e27d180ef3',1,'copy.h']]], - ['cos_82',['Cos',['../classmlx_1_1core_1_1_cos.html#a2acb9fcf0901462189c476756fd99995',1,'mlx::core::Cos']]], - ['cos_83',['cos',['../namespacepocketfft_1_1detail.html#a499c1e8b7d79a5272af024f46c63ff9d',1,'pocketfft::detail::cos()'],['../namespacemetal.html#a2fa4778a6fe2fa43253ea724e5a608a3',1,'metal::cos()'],['../namespacemetal_1_1fast.html#a75b6bb32fa3870eda46a7bfc9f481f88',1,'metal::fast::cos()'],['../namespacemetal_1_1precise.html#ac4941f62e7d8ab9d7cabbd967aa9f220',1,'metal::precise::cos()'],['../group__ops.html#ga39dfdf72b556012aa35ff27a94116e74',1,'mlx::core::cos()']]], - ['cosh_84',['Cosh',['../classmlx_1_1core_1_1_cosh.html#a44e8ac2e09a55ec32e9dc6641eedc8f1',1,'mlx::core::Cosh']]], - ['cosh_85',['cosh',['../namespacemetal.html#a8a68a88cc110830d057dbd71431b93c0',1,'metal::cosh()'],['../namespacemetal_1_1fast.html#a31544ad9de28012a4ddda86e3966a77e',1,'metal::fast::cosh()'],['../namespacemetal_1_1precise.html#a72d86d508300a9b58f4ccbbe70da4fbc',1,'metal::precise::cosh()'],['../group__ops.html#ga2181b71cda88007a3092be4795ff0715',1,'mlx::core::cosh()']]], - ['cospi_86',['cospi',['../namespacemetal.html#a5c2f37939ad705ddea4409d3bedb8ce1',1,'metal::cospi()'],['../namespacemetal_1_1fast.html#a9906b41f75319b384ffb570cc94d67ce',1,'metal::fast::cospi()'],['../namespacemetal_1_1precise.html#a2392b78bd196efdbbac65901c4ab20e7',1,'metal::precise::cospi()']]], - ['cost_5fguess_87',['cost_guess',['../structpocketfft_1_1detail_1_1util.html#ad3d874bc3fb0048df2270779a15d4bd0',1,'pocketfft::detail::util']]], - ['count_5fdown_88',['count_down',['../classpocketfft_1_1detail_1_1threading_1_1latch.html#a81d6597189b40410e35f3cd653fd1342',1,'pocketfft::detail::threading::latch']]], - ['cross_89',['cross',['../namespacemlx_1_1core_1_1linalg.html#abcda3fbda45183c21e7f27aa0dde64e6',1,'mlx::core::linalg']]], - ['cummax_90',['cummax',['../group__ops.html#gaee37cac8476e8f8d666bcded5bc59143',1,'mlx::core']]], - ['cummin_91',['cummin',['../group__ops.html#ga19c1bf6929fe8d66b9cd408946aea6a8',1,'mlx::core']]], - ['cumprod_92',['cumprod',['../group__ops.html#ga0d71dfbc14ef3ed564b0c5ee26af680f',1,'mlx::core']]], - ['cumsum_93',['cumsum',['../group__ops.html#gaddc825a5c173e195ab0fda83ad630420',1,'mlx::core']]], - ['custom_94',['Custom',['../classmlx_1_1core_1_1fast_1_1_custom.html#a4186fea23f7156c38960426821fca313',1,'mlx::core::fast::Custom']]], - ['custom_5ffunction_95',['custom_function',['../namespacemlx_1_1core.html#a8d3ca5fbaecdb995660c24cde5aeebaf',1,'mlx::core']]], - ['custom_5fvjp_96',['custom_vjp',['../namespacemlx_1_1core.html#a9290596250fa308df4c69b44483bb8aa',1,'mlx::core']]], - ['customkernel_97',['CustomKernel',['../classmlx_1_1core_1_1fast_1_1_custom_kernel.html#a954893e07f0d36715b4e1e414b6f2153',1,'mlx::core::fast::CustomKernel']]], - ['customtransforms_98',['CustomTransforms',['../classmlx_1_1core_1_1_custom_transforms.html#ab52abadb9c6f6db83d087c7b751be488',1,'mlx::core::CustomTransforms']]] + ['col_5freduce_5f2pass_18',['col_reduce_2pass',['../reduce__col_8h.html#a0e92fc74eeaa8ee2ceb83bafc6eb1d7d',1,'reduce_col.h']]], + ['col_5freduce_5flongcolumn_19',['col_reduce_longcolumn',['../reduce__col_8h.html#a5b4f4c4c247ad341ff8d31dcbbbce0eb',1,'reduce_col.h']]], + ['col_5freduce_5flooped_20',['col_reduce_looped',['../reduce__col_8h.html#a11bfc6112ae2386ac03f5ea7b7d93385',1,'reduce_col.h']]], + ['col_5freduce_5fsmall_21',['col_reduce_small',['../reduce__col_8h.html#a7c378443a2b6f4d9210db8a21a9ac4f5',1,'reduce_col.h']]], + ['collapse_5fcontiguous_5fdims_22',['collapse_contiguous_dims',['../namespacemlx_1_1core.html#a38fe6ec5220d13d96c7dad7556d2b613',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< std::vector< int64_t > > &strides, int64_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#af2895f9b0083efd8221275eb8cadccbe',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< std::vector< size_t > > &strides, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a90e2b6edc0fe82230cb93f5ea39febb4',1,'mlx::core::collapse_contiguous_dims(const std::vector< array > &xs, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#ac813412cce77fc1340dcfefc6e099276',1,'mlx::core::collapse_contiguous_dims(Arrays &&... xs)'],['../namespacemlx_1_1core.html#aab3cc7f3808934ae0727b920eba231bd',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< int64_t > &strides, int64_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a1e0cbcf109d32794ffc8efc7302ba9b0',1,'mlx::core::collapse_contiguous_dims(const std::vector< int > &shape, const std::vector< size_t > &strides, size_t size_cap=std::numeric_limits< int32_t >::max())'],['../namespacemlx_1_1core.html#a4ee50bfb240512d0c0ce151dfe2c74ef',1,'mlx::core::collapse_contiguous_dims(const array &a, size_t size_cap=std::numeric_limits< int32_t >::max())']]], + ['commandencoder_23',['CommandEncoder',['../structmlx_1_1core_1_1metal_1_1_command_encoder.html#a2334774486f447213ee997e55c2e52a3',1,'mlx::core::metal::CommandEncoder::CommandEncoder(MTL::CommandBuffer *cbuf)'],['../structmlx_1_1core_1_1metal_1_1_command_encoder.html#ac68ca977b5bde5434284ce7979647f14',1,'mlx::core::metal::CommandEncoder::CommandEncoder(const CommandEncoder &)=delete']]], + ['commit_5fcommand_5fbuffer_24',['commit_command_buffer',['../classmlx_1_1core_1_1metal_1_1_device.html#a95248f1387824067fd4fed23ace5ac0c',1,'mlx::core::metal::Device']]], + ['communication_5fstream_25',['communication_stream',['../namespacemlx_1_1core_1_1distributed_1_1detail.html#ac3612edf0e0e18c1e4ba0ce7c6e35cd6',1,'mlx::core::distributed::detail']]], + ['compile_26',['compile',['../namespacemlx_1_1core.html#a3ac798e65e59fe10b7fb5c522efce782',1,'mlx::core::compile()'],['../namespacemlx_1_1core_1_1detail.html#ac3b7b09892ff7290d5f3ef26cb444329',1,'mlx::core::detail::compile(const std::function< std::vector< array >(const std::vector< array > &)> &fun, std::uintptr_t fun_id, bool shapeless=false, std::vector< uint64_t > constants={})']]], + ['compile_5favailable_5ffor_5fdevice_27',['compile_available_for_device',['../namespacemlx_1_1core_1_1detail.html#aeeff2ba6ec3d9d4ed090de6d2681dbc2',1,'mlx::core::detail']]], + ['compile_5fclear_5fcache_28',['compile_clear_cache',['../namespacemlx_1_1core_1_1detail.html#a3fb927c209b946aefebb195993fbe4cf',1,'mlx::core::detail']]], + ['compile_5ferase_29',['compile_erase',['../namespacemlx_1_1core_1_1detail.html#a69eb76a14f845ca000f1ccb2edda0175',1,'mlx::core::detail']]], + ['compiled_30',['Compiled',['../classmlx_1_1core_1_1_compiled.html#a2d8cefff835c419a48a077d306b8e051',1,'mlx::core::Compiled']]], + ['compiled_5fallocate_5foutputs_31',['compiled_allocate_outputs',['../namespacemlx_1_1core.html#ab8c3c4fc05745f586de922c8266f4fce',1,'mlx::core']]], + ['compiled_5fcheck_5fcontiguity_32',['compiled_check_contiguity',['../namespacemlx_1_1core.html#a3b900ab319948c5a01a3ecd30a709027',1,'mlx::core']]], + ['complex128_5ft_33',['complex128_t',['../structmlx_1_1core_1_1complex128__t.html#aa15d0b805f8790f7c7b76fc7b9d677e0',1,'mlx::core::complex128_t::complex128_t(double v, double u)'],['../structmlx_1_1core_1_1complex128__t.html#abf2842253b874f9f13f39ea68a89e5b6',1,'mlx::core::complex128_t::complex128_t(std::complex< double > v)'],['../structmlx_1_1core_1_1complex128__t.html#a526fba96d7e815360cb4226af085a1bf',1,'mlx::core::complex128_t::complex128_t(T x)']]], + ['complex64_5ft_34',['complex64_t',['../structcomplex64__t.html#adbd392a5e92d31997380ad0a38be4be8',1,'complex64_t::complex64_t(float real, float imag)'],['../structcomplex64__t.html#a29782289bb90d6294099667b86509cd3',1,'complex64_t::complex64_t()'],['../structcomplex64__t.html#a905b048d70eb8d748a62454268242291',1,'complex64_t::complex64_t() threadgroup'],['../structcomplex64__t.html#a33a2452eb33b5ed53655773539c357a5',1,'complex64_t::complex64_t(T x) thread'],['../structcomplex64__t.html#a89b65ace8588b7bf215355f705eb23d9',1,'complex64_t::complex64_t(T x) threadgroup'],['../structcomplex64__t.html#ac81b486f642fb3b26c5d659917bdbcd0',1,'complex64_t::complex64_t(T x) device'],['../structcomplex64__t.html#a0a27a41206400f1e62b60ceb56960c93',1,'complex64_t::complex64_t(T x) const ant'],['../structmlx_1_1core_1_1complex64__t.html#a697cc973ae27d63c8e00d830e780bd8c',1,'mlx::core::complex64_t::complex64_t(float v, float u)'],['../structmlx_1_1core_1_1complex64__t.html#ae065e39938f9c4374b4116f4c67d4d09',1,'mlx::core::complex64_t::complex64_t(std::complex< float > v)'],['../structmlx_1_1core_1_1complex64__t.html#a2232cbbe591a9d2bc228cb23fac38b50',1,'mlx::core::complex64_t::complex64_t(T x)']]], + ['complex_5fmul_35',['complex_mul',['../radix_8h.html#a5bfc53b531214c9ce277bebc18aa67d6',1,'radix.h']]], + ['complex_5fmul_5fconj_36',['complex_mul_conj',['../radix_8h.html#a0e2dfd3d1dda09f47ccc64eec35629f3',1,'radix.h']]], + ['compute_5fstrided_5findices_37',['compute_strided_indices',['../struct_read_writer.html#a7c903fbb8b85a856ba5564d7df537cdf',1,'ReadWriter']]], + ['concatenate_38',['Concatenate',['../classmlx_1_1core_1_1_concatenate.html#acff07853de2d31faeec7c4ca40ce0888',1,'mlx::core::Concatenate']]], + ['concatenate_39',['concatenate',['../group__ops.html#gabdc36fa65697d0361c8d67495de77129',1,'mlx::core::concatenate(const std::vector< array > &arrays, int axis, StreamOrDevice s={})'],['../group__ops.html#gaa95c34ca3a8877f2c50cb60e7fa312b8',1,'mlx::core::concatenate(const std::vector< array > &arrays, StreamOrDevice s={})']]], + ['concatenate_5fgpu_40',['concatenate_gpu',['../namespacemlx_1_1core.html#a050299d0d366ca5c9d09d1004dcc3e7d',1,'mlx::core']]], + ['concurrentcontext_41',['ConcurrentContext',['../structmlx_1_1core_1_1metal_1_1_command_encoder_1_1_concurrent_context.html#aee044d7729739c96e845823f9ecc5174',1,'mlx::core::metal::CommandEncoder::ConcurrentContext']]], + ['conj_42',['conj',['../namespacepocketfft_1_1detail.html#a66d79051d502046a9b9f103e744dbad3',1,'pocketfft::detail']]], + ['conjugate_43',['Conjugate',['../classmlx_1_1core_1_1_conjugate.html#a627f9e6a8729fb3ffb3ca3228d007c87',1,'mlx::core::Conjugate']]], + ['conjugate_44',['conjugate',['../group__ops.html#ga5b596906bf8cdc8d97ed6ddc9aeb4c23',1,'mlx::core']]], + ['contiguous_5fscan_45',['contiguous_scan',['../scan_8h.html#a60d279b9add7d56639bb209408f09d79',1,'scan.h']]], + ['contiguousiterator_46',['ContiguousIterator',['../structmlx_1_1core_1_1_contiguous_iterator.html#a68794af4a442d3d8ac4647817af8e1f6',1,'mlx::core::ContiguousIterator::ContiguousIterator()'],['../structmlx_1_1core_1_1_contiguous_iterator.html#a6cb378408b6f546eeb6ade1a4faafe3c',1,'mlx::core::ContiguousIterator::ContiguousIterator(const array &a)'],['../structmlx_1_1core_1_1_contiguous_iterator.html#a16bdacb53f65b7284068cd49d4cba292',1,'mlx::core::ContiguousIterator::ContiguousIterator(const std::vector< int > &shape, const std::vector< StrideT > &strides, int dims)']]], + ['conv_47',['conv',['../namespacemlx_1_1core_1_1metal.html#ab1704e853394c725668c06752ebb5c24',1,'mlx::core::metal']]], + ['conv1d_48',['conv1d',['../group__ops.html#ga30d47e08093c03a3676f235f9f559411',1,'mlx::core']]], + ['conv2d_49',['conv2d',['../group__ops.html#ga73b02833229678786e7f302d458d5a83',1,'mlx::core']]], + ['conv2dinputblockloadergeneral_50',['Conv2DInputBlockLoaderGeneral',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_general.html#a1d83af561a483432bf8dcb42e734b23b',1,'mlx::steel::Conv2DInputBlockLoaderGeneral']]], + ['conv2dinputblockloaderlargefilter_51',['Conv2DInputBlockLoaderLargeFilter',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_large_filter.html#a8755116a535539744e4947bc69f9c50f',1,'mlx::steel::Conv2DInputBlockLoaderLargeFilter']]], + ['conv2dinputblockloadersmallchannels_52',['Conv2DInputBlockLoaderSmallChannels',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_channels.html#ab9fd3fdeab94470dde3326f1dd5c455a',1,'mlx::steel::Conv2DInputBlockLoaderSmallChannels']]], + ['conv2dinputblockloadersmallfilter_53',['Conv2DInputBlockLoaderSmallFilter',['../structmlx_1_1steel_1_1_conv2_d_input_block_loader_small_filter.html#a0a2cbf57c51cd928722e3f06aafcf933',1,'mlx::steel::Conv2DInputBlockLoaderSmallFilter']]], + ['conv2dweightblockloader_54',['Conv2DWeightBlockLoader',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader.html#a9a7dca3512b64cffb6eac305d795831c',1,'mlx::steel::Conv2DWeightBlockLoader']]], + ['conv2dweightblockloadergeneral_55',['Conv2DWeightBlockLoaderGeneral',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_general.html#ad0550fabbdc9297559381a5b488e9af1',1,'mlx::steel::Conv2DWeightBlockLoaderGeneral']]], + ['conv2dweightblockloadersmallchannels_56',['Conv2DWeightBlockLoaderSmallChannels',['../structmlx_1_1steel_1_1_conv2_d_weight_block_loader_small_channels.html#ae1806ea1c19713819dee83a38ab35fa6',1,'mlx::steel::Conv2DWeightBlockLoaderSmallChannels']]], + ['conv3d_57',['conv3d',['../group__ops.html#ga6e9907d2f14dc4803e4306b3dbc4b3ca',1,'mlx::core']]], + ['conv_5fgeneral_58',['conv_general',['../group__ops.html#ga2236e5dfc7e52e28abf6c21675d0a51e',1,'mlx::core::conv_general(array input, array weight, std::vector< int > stride={}, std::vector< int > padding_lo={}, std::vector< int > padding_hi={}, std::vector< int > kernel_dilation={}, std::vector< int > input_dilation={}, int groups=1, bool flip=false, StreamOrDevice s={})'],['../group__ops.html#gab59f89942cd1efaadffe9e8762e3c99d',1,'mlx::core::conv_general(const array &input, const array &weight, std::vector< int > stride={}, std::vector< int > padding={}, std::vector< int > kernel_dilation={}, std::vector< int > input_dilation={}, int groups=1, bool flip=false, StreamOrDevice s={})']]], + ['conv_5ftranspose1d_59',['conv_transpose1d',['../group__ops.html#gaa30bf1adcd78d1c2595d07b215731714',1,'mlx::core']]], + ['conv_5ftranspose2d_60',['conv_transpose2d',['../group__ops.html#gaebb59971cb9bc45005dc1d398e4f0a3d',1,'mlx::core']]], + ['conv_5ftranspose3d_61',['conv_transpose3d',['../group__ops.html#ga8db814da631d9cd32a8d6563bf4ac530',1,'mlx::core']]], + ['convolution_62',['Convolution',['../classmlx_1_1core_1_1_convolution.html#a6f1de77b719bb13217b0d8c64cabb8ef',1,'mlx::core::Convolution']]], + ['copy_63',['Copy',['../classmlx_1_1core_1_1_copy.html#a6243e044af119105ffaaed7d405cd584',1,'mlx::core::Copy']]], + ['copy_64',['copy',['../namespacemlx_1_1core.html#a479648542a2bea151b947b18f0e79dd2',1,'mlx::core::copy()'],['../namespacemlx_1_1core_1_1metal.html#aa215e631e2680f04a591b88d91571719',1,'mlx::core::metal::copy()'],['../group__ops.html#gae306e93af12f774bd80bad6c231b09d6',1,'mlx::core::copy()']]], + ['copy_5fg_65',['copy_g',['../metal_2kernels_2copy_8h.html#a778ce2dbfbaa23b24bd5efbe68448c36',1,'copy.h']]], + ['copy_5fg_5fnd1_66',['copy_g_nd1',['../metal_2kernels_2copy_8h.html#aba4530a7db6a61ca36f50e4f5e58fb77',1,'copy.h']]], + ['copy_5fg_5fnd2_67',['copy_g_nd2',['../metal_2kernels_2copy_8h.html#aee678c7c31119f3e609685589f37490c',1,'copy.h']]], + ['copy_5fg_5fnd3_68',['copy_g_nd3',['../metal_2kernels_2copy_8h.html#a821f8f3f3891159a295c66fc25aed1ff',1,'copy.h']]], + ['copy_5fgg_69',['copy_gg',['../metal_2kernels_2copy_8h.html#a1e39c2683eeaf05955e7619fbd34aea5',1,'copy.h']]], + ['copy_5fgg_5fnd1_70',['copy_gg_nd1',['../metal_2kernels_2copy_8h.html#a3278d9c999718bee3ccbe2922f501bf1',1,'copy.h']]], + ['copy_5fgg_5fnd2_71',['copy_gg_nd2',['../metal_2kernels_2copy_8h.html#a3e2d3cc7f34f56170409b6735f51a950',1,'copy.h']]], + ['copy_5fgg_5fnd3_72',['copy_gg_nd3',['../metal_2kernels_2copy_8h.html#a59f43b5bffed936d7559ceb06a10aabd',1,'copy.h']]], + ['copy_5fgpu_73',['copy_gpu',['../namespacemlx_1_1core.html#addaa46a13ac2deb1d9ce621338320e0e',1,'mlx::core::copy_gpu(const array &src, array &out, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#a6a6f4e46c8fc44fdc74c50ace02bcf38',1,'mlx::core::copy_gpu(const array &src, array &out, CopyType ctype)']]], + ['copy_5fgpu_5finplace_74',['copy_gpu_inplace',['../namespacemlx_1_1core.html#a69e30f5d30a6d72ac0ffe4886f24b7ba',1,'mlx::core::copy_gpu_inplace(const array &in, array &out, const std::vector< int > &data_shape, const std::vector< stride_t > &i_strides, const std::vector< stride_t > &o_strides, int64_t i_offset, int64_t o_offset, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#a8e1ccb0ed9387b0a789311d9f8964803',1,'mlx::core::copy_gpu_inplace(const array &src, array &out, CopyType ctype, const Stream &s)'],['../namespacemlx_1_1core.html#ae55b801b09ccf55cba96278163a9b1ef',1,'mlx::core::copy_gpu_inplace(const array &in, array &out, const std::vector< int64_t > &istride, int64_t ioffset, CopyType ctype, const Stream &s)']]], + ['copy_5fhartley_75',['copy_hartley',['../namespacepocketfft_1_1detail.html#abac3fcc8ce83800d228774f64c28d4c3',1,'pocketfft::detail::copy_hartley(const multi_iter< vlen > &it, const vtype_t< T > *src, ndarr< T > &dst)'],['../namespacepocketfft_1_1detail.html#ae7b44d2773d9d06a9787aff01d66b3ed',1,'pocketfft::detail::copy_hartley(const multi_iter< vlen > &it, const T *src, ndarr< T > &dst)']]], + ['copy_5finplace_76',['copy_inplace',['../namespacemlx_1_1core.html#a98495894a796b2cc6d022e7a03432c64',1,'mlx::core::copy_inplace(const array &src, array &dst, CopyType ctype)'],['../namespacemlx_1_1core.html#aad636e2d0b2f882cadd1b438f4daa9ed',1,'mlx::core::copy_inplace(const array &src, array &dst, const std::vector< int > &data_shape, const std::vector< stride_t > &i_strides, const std::vector< stride_t > &o_strides, int64_t i_offset, int64_t o_offset, CopyType ctype)']]], + ['copy_5finput_77',['copy_input',['../namespacepocketfft_1_1detail.html#aff05be3064743c1143b19318ab12ad4a',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< cmplx< T > > &src, cmplx< vtype_t< T > > *dst)'],['../namespacepocketfft_1_1detail.html#a30fc708f9d8f9cfa74194925c7863c0a',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< T > &src, vtype_t< T > *dst)'],['../namespacepocketfft_1_1detail.html#a3387bd35f237870e42b8461769e6aec4',1,'pocketfft::detail::copy_input(const multi_iter< vlen > &it, const cndarr< T > &src, T *dst)']]], + ['copy_5foutput_78',['copy_output',['../namespacepocketfft_1_1detail.html#a1523a037300a8da05db210b802d9cb0e',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const cmplx< vtype_t< T > > *src, ndarr< cmplx< T > > &dst)'],['../namespacepocketfft_1_1detail.html#a21980853aca4d92ed06e3dcffe7ef660',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const vtype_t< T > *src, ndarr< T > &dst)'],['../namespacepocketfft_1_1detail.html#a310481c334e46674710ba794ad7403c0',1,'pocketfft::detail::copy_output(const multi_iter< vlen > &it, const T *src, ndarr< T > &dst)']]], + ['copy_5fs_79',['copy_s',['../metal_2kernels_2copy_8h.html#aef09f9b9475345b1bba121d037d222ea',1,'copy.h']]], + ['copy_5fs2_80',['copy_s2',['../metal_2kernels_2copy_8h.html#a8023e9335cc5334847a8d315042be3a3',1,'copy.h']]], + ['copy_5fshared_5fbuffer_81',['copy_shared_buffer',['../classmlx_1_1core_1_1array.html#a28df7a333d90a311c49bc4bce7a1ad6d',1,'mlx::core::array::copy_shared_buffer(const array &other, const std::vector< size_t > &strides, Flags flags, size_t data_size, size_t offset=0)'],['../classmlx_1_1core_1_1array.html#a92974c656c35a972ad241f80584bbd29',1,'mlx::core::array::copy_shared_buffer(const array &other)']]], + ['copy_5fv_82',['copy_v',['../metal_2kernels_2copy_8h.html#ae26a13e0c8e6c15f7b10078e65970659',1,'copy.h']]], + ['copy_5fv2_83',['copy_v2',['../metal_2kernels_2copy_8h.html#aee14a5326f53d9b30b0b38e27d180ef3',1,'copy.h']]], + ['cos_84',['Cos',['../classmlx_1_1core_1_1_cos.html#a2acb9fcf0901462189c476756fd99995',1,'mlx::core::Cos']]], + ['cos_85',['cos',['../namespacepocketfft_1_1detail.html#a499c1e8b7d79a5272af024f46c63ff9d',1,'pocketfft::detail::cos()'],['../namespacemetal.html#a2fa4778a6fe2fa43253ea724e5a608a3',1,'metal::cos()'],['../namespacemetal_1_1fast.html#a75b6bb32fa3870eda46a7bfc9f481f88',1,'metal::fast::cos()'],['../namespacemetal_1_1precise.html#ac4941f62e7d8ab9d7cabbd967aa9f220',1,'metal::precise::cos()'],['../group__ops.html#ga39dfdf72b556012aa35ff27a94116e74',1,'mlx::core::cos()']]], + ['cosh_86',['Cosh',['../classmlx_1_1core_1_1_cosh.html#a44e8ac2e09a55ec32e9dc6641eedc8f1',1,'mlx::core::Cosh']]], + ['cosh_87',['cosh',['../namespacemetal.html#a8a68a88cc110830d057dbd71431b93c0',1,'metal::cosh()'],['../namespacemetal_1_1fast.html#a31544ad9de28012a4ddda86e3966a77e',1,'metal::fast::cosh()'],['../namespacemetal_1_1precise.html#a72d86d508300a9b58f4ccbbe70da4fbc',1,'metal::precise::cosh()'],['../group__ops.html#ga2181b71cda88007a3092be4795ff0715',1,'mlx::core::cosh()']]], + ['cospi_88',['cospi',['../namespacemetal.html#a5c2f37939ad705ddea4409d3bedb8ce1',1,'metal::cospi()'],['../namespacemetal_1_1fast.html#a9906b41f75319b384ffb570cc94d67ce',1,'metal::fast::cospi()'],['../namespacemetal_1_1precise.html#a2392b78bd196efdbbac65901c4ab20e7',1,'metal::precise::cospi()']]], + ['cost_5fguess_89',['cost_guess',['../structpocketfft_1_1detail_1_1util.html#ad3d874bc3fb0048df2270779a15d4bd0',1,'pocketfft::detail::util']]], + ['count_5fdown_90',['count_down',['../classpocketfft_1_1detail_1_1threading_1_1latch.html#a81d6597189b40410e35f3cd653fd1342',1,'pocketfft::detail::threading::latch']]], + ['cross_91',['cross',['../namespacemlx_1_1core_1_1linalg.html#abcda3fbda45183c21e7f27aa0dde64e6',1,'mlx::core::linalg']]], + ['cummax_92',['cummax',['../group__ops.html#gaee37cac8476e8f8d666bcded5bc59143',1,'mlx::core']]], + ['cummin_93',['cummin',['../group__ops.html#ga19c1bf6929fe8d66b9cd408946aea6a8',1,'mlx::core']]], + ['cumprod_94',['cumprod',['../group__ops.html#ga0d71dfbc14ef3ed564b0c5ee26af680f',1,'mlx::core']]], + ['cumsum_95',['cumsum',['../group__ops.html#gaddc825a5c173e195ab0fda83ad630420',1,'mlx::core']]], + ['custom_96',['Custom',['../classmlx_1_1core_1_1fast_1_1_custom.html#a4186fea23f7156c38960426821fca313',1,'mlx::core::fast::Custom']]], + ['custom_5ffunction_97',['custom_function',['../namespacemlx_1_1core.html#a8d3ca5fbaecdb995660c24cde5aeebaf',1,'mlx::core']]], + ['custom_5fvjp_98',['custom_vjp',['../namespacemlx_1_1core.html#a9290596250fa308df4c69b44483bb8aa',1,'mlx::core']]], + ['customkernel_99',['CustomKernel',['../classmlx_1_1core_1_1fast_1_1_custom_kernel.html#a954893e07f0d36715b4e1e414b6f2153',1,'mlx::core::fast::CustomKernel']]], + ['customtransforms_100',['CustomTransforms',['../classmlx_1_1core_1_1_custom_transforms.html#ab52abadb9c6f6db83d087c7b751be488',1,'mlx::core::CustomTransforms']]] ]; diff --git a/docs/build/html/search/functions_7.js b/docs/build/html/search/functions_7.js index 2b5a027d5..91a1bdfb7 100644 --- a/docs/build/html/search/functions_7.js +++ b/docs/build/html/search/functions_7.js @@ -44,7 +44,7 @@ var searchData= ['get_5fpool_41',['get_pool',['../namespacepocketfft_1_1detail_1_1threading.html#a7ec2b3f99232bd0f15f7b022c59d139a',1,'pocketfft::detail::threading']]], ['get_5fprimitive_5fstring_42',['get_primitive_string',['../namespacemlx_1_1core.html#ad4be35b310a252edd80d9cf04f094a60',1,'mlx::core']]], ['get_5fquantized_5fkernel_43',['get_quantized_kernel',['../namespacemlx_1_1core.html#aa3faeae5378bfaafe3ce3432a051e43e',1,'mlx::core']]], - ['get_5freduce_5finit_5fkernel_44',['get_reduce_init_kernel',['../namespacemlx_1_1core.html#a51c4bb09230348bd0252e22bfdc9bc89',1,'mlx::core']]], + ['get_5freduce_5finit_5fkernel_44',['get_reduce_init_kernel',['../namespacemlx_1_1core.html#a3bd386cb6db09f636963ce66ceaf8647',1,'mlx::core']]], ['get_5freduce_5fkernel_45',['get_reduce_kernel',['../namespacemlx_1_1core.html#a7aa91fcfe8b9caa42d60a957f11bfe6b',1,'mlx::core']]], ['get_5freduction_5fplan_46',['get_reduction_plan',['../namespacemlx_1_1core.html#ac97b5a6f009ca3d99854ce9512c20dba',1,'mlx::core']]], ['get_5fscan_5fkernel_47',['get_scan_kernel',['../namespacemlx_1_1core.html#aeefaff208444d3fa61ecc0946fe1de5f',1,'mlx::core']]], diff --git a/docs/build/html/search/functions_d.js b/docs/build/html/search/functions_d.js index 5e2da1c4d..7967de990 100644 --- a/docs/build/html/search/functions_d.js +++ b/docs/build/html/search/functions_d.js @@ -14,39 +14,40 @@ var searchData= ['max3_11',['max3',['../namespacemetal.html#a00f9c0ad66d969794614f56912eed9c9',1,'metal::max3()'],['../namespacemetal_1_1fast.html#a6fc2cf18ffa8149561864c86dba0f803',1,'metal::fast::max3()'],['../namespacemetal_1_1precise.html#ac490e8614ebd2c9343af1ae6c0d4e82c',1,'metal::precise::max3()']]], ['maximum_12',['Maximum',['../classmlx_1_1core_1_1_maximum.html#a28389307e385efe1b2955b86b115e816',1,'mlx::core::Maximum']]], ['maximum_13',['maximum',['../group__ops.html#ga7ade2ea305e2e4219c3609443fb5db8d',1,'mlx::core']]], - ['mb_5fblock_5fmerge_14',['mb_block_merge',['../sort_8h.html#ab381cd57f344bc7304ab580bfdc78807',1,'sort.h']]], - ['mb_5fblock_5fpartition_15',['mb_block_partition',['../sort_8h.html#a32cbe4163b8b0f5cb2c97b256119a4b2',1,'sort.h']]], - ['mb_5fblock_5fsort_16',['mb_block_sort',['../sort_8h.html#aa48ff1aff1e9dc1301b6781aa0721d6b',1,'sort.h']]], - ['mean_17',['mean',['../group__ops.html#gade46e768fd46b8b640eb16f26abeecef',1,'mlx::core::mean(const array &a, bool keepdims, StreamOrDevice s={})'],['../group__ops.html#ga52b59fdd8e8430538e564f5bbcfa31e6',1,'mlx::core::mean(const array &a, StreamOrDevice s={})'],['../group__ops.html#ga066161f3d3e395a1d76c638cb680d444',1,'mlx::core::mean(const array &a, const std::vector< int > &axes, bool keepdims=false, StreamOrDevice s={})'],['../group__ops.html#ga45fba73eab0e3b6e128ed3ce2f43a5da',1,'mlx::core::mean(const array &a, int axis, bool keepdims=false, StreamOrDevice s={})']]], - ['median3_18',['median3',['../namespacemetal.html#aa3ff49457ce3c93fc1c0897fd1525157',1,'metal::median3()'],['../namespacemetal_1_1fast.html#a742b55f1e4369921ee7f60d70185bfbc',1,'metal::fast::median3()'],['../namespacemetal_1_1precise.html#a14555ff99c4388493fec48e070144ae2',1,'metal::precise::median3()']]], - ['merge_5fpartition_19',['merge_partition',['../struct_block_merge_sort.html#ab2300cbecb23f3433bad888924c831ca',1,'BlockMergeSort::merge_partition()'],['../struct_kernel_multi_block_merge_sort.html#ab15895b4233aba0e279cc44a07a201fe',1,'KernelMultiBlockMergeSort::merge_partition()']]], - ['merge_5fstep_20',['merge_step',['../struct_block_merge_sort.html#ab65f190edf1851b37c39ad49ce99a43c',1,'BlockMergeSort']]], - ['meshgrid_21',['meshgrid',['../group__ops.html#ga577c911618575314de63d1060656a26e',1,'mlx::core']]], - ['metal_5fkernel_22',['metal_kernel',['../namespacemlx_1_1core_1_1fast.html#ab16436b465dc10ce472193d541d8426e',1,'mlx::core::fast']]], - ['min_23',['min',['../structmetal_1_1__numeric__limits__impl_3_01bfloat16__t_01_4.html#adaed80031f5ca0ff69d30ec4c5d0c98f',1,'metal::_numeric_limits_impl< bfloat16_t >::min()'],['../namespacemetal.html#a6653b28c9473087141eddce39878d4d3',1,'metal::min()'],['../namespacemetal_1_1fast.html#a3e958e56a4712687c381a0b64d123e61',1,'metal::fast::min()'],['../namespacemetal_1_1precise.html#afed0da2f7df3505b5dffa2389c3cb36e',1,'metal::precise::min()'],['../group__ops.html#gab27599802617a4c8f9964ab5f4ffee12',1,'mlx::core::min(const array &a, bool keepdims, StreamOrDevice s={})'],['../group__ops.html#ga0140b91e9cdfc3fef0da8e332f65a9e8',1,'mlx::core::min(const array &a, StreamOrDevice s={})'],['../group__ops.html#ga6efb83cd46436678c8f8c4af15cc00f5',1,'mlx::core::min(const array &a, const std::vector< int > &axes, bool keepdims=false, StreamOrDevice s={})'],['../group__ops.html#ga36fa315eef677f4143868f552cd26d03',1,'mlx::core::min(const array &a, int axis, bool keepdims=false, StreamOrDevice s={})']]], - ['min3_24',['min3',['../namespacemetal.html#a005510c8c0f964ce2b8aad3ba76a7a3f',1,'metal::min3()'],['../namespacemetal_1_1fast.html#a606a4c1b34ce05ea89ca5af81724036f',1,'metal::fast::min3()'],['../namespacemetal_1_1precise.html#a4d37ce31c3549ca4772a4ee29798e231',1,'metal::precise::min3()']]], - ['minimum_25',['Minimum',['../classmlx_1_1core_1_1_minimum.html#ab0f2ce17108df44b82cff68886b0f6f5',1,'mlx::core::Minimum']]], - ['minimum_26',['minimum',['../group__ops.html#ga49ba00c090f81f331c91b0c97040bce0',1,'mlx::core']]], - ['mlx_5fatomic_5fcompare_5fexchange_5fweak_5fexplicit_27',['mlx_atomic_compare_exchange_weak_explicit',['../atomic_8h.html#ad7f32327ff66354cfa2f0cfdac79316f',1,'mlx_atomic_compare_exchange_weak_explicit(device mlx_atomic< T > *object, thread T *expected, T val, size_t offset): atomic.h'],['../atomic_8h.html#aa8f47b2e9b95d4b00ad51f08b070deb5',1,'mlx_atomic_compare_exchange_weak_explicit(device mlx_atomic< T > *object, thread uint *expected, uint val, size_t offset): atomic.h']]], - ['mlx_5fatomic_5ffetch_5fadd_5fexplicit_28',['mlx_atomic_fetch_add_explicit',['../atomic_8h.html#aad448d9e06e001700b65ca8317216a3b',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fand_5fexplicit_29',['mlx_atomic_fetch_and_explicit',['../atomic_8h.html#a253e3c870c0ddc7c28ab2f6ca2c3eae5',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fmax_5fexplicit_30',['mlx_atomic_fetch_max_explicit',['../atomic_8h.html#ac480f2b459a8ad9095cee353e152d00c',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fmax_5fexplicit_3c_20float_20_3e_31',['mlx_atomic_fetch_max_explicit< float >',['../atomic_8h.html#a1dce2abfa16417122c4d2bf261129ae4',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fmin_5fexplicit_32',['mlx_atomic_fetch_min_explicit',['../atomic_8h.html#a2ec33dca0039bd944d73d1c2b378cc19',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fmin_5fexplicit_3c_20float_20_3e_33',['mlx_atomic_fetch_min_explicit< float >',['../atomic_8h.html#ab7d1dc49f319f239b7ee0b7c72976dd0',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5fmul_5fexplicit_34',['mlx_atomic_fetch_mul_explicit',['../atomic_8h.html#adfdbea60436f14f1af9ce36e2a0a77a3',1,'atomic.h']]], - ['mlx_5fatomic_5ffetch_5for_5fexplicit_35',['mlx_atomic_fetch_or_explicit',['../atomic_8h.html#ab7391f197001471e4788312bdb6ab37a',1,'atomic.h']]], - ['mlx_5fatomic_5fload_5fexplicit_36',['mlx_atomic_load_explicit',['../atomic_8h.html#a253a4e8c2c5768a069e2791b627dfc99',1,'atomic.h']]], - ['mlx_5fatomic_5fstore_5fexplicit_37',['mlx_atomic_store_explicit',['../atomic_8h.html#a0ae453140b0819a4c02f265334de98c0',1,'atomic.h']]], - ['mma_38',['mma',['../structmlx_1_1steel_1_1_base_m_m_a_frag_3_01_t_00_018_00_018_01_4.html#a8028512f5a3d2b6acaf966be529627a3',1,'mlx::steel::BaseMMAFrag< T, 8, 8 >::mma(thread frag_type &D, thread frag_type &A, thread frag_type &B, thread frag_type &C)'],['../structmlx_1_1steel_1_1_base_m_m_a_frag_3_01_t_00_018_00_018_01_4.html#a1868f57d57c8adedab2c58492ec76946',1,'mlx::steel::BaseMMAFrag< T, 8, 8 >::mma(thread mat_type &D, thread mat_type &A, thread mat_type &B, thread mat_type &C)'],['../structmlx_1_1steel_1_1_block_m_m_a.html#a6a2c2a6d5e767d52c41b42a9d36086b0',1,'mlx::steel::BlockMMA::mma()']]], - ['mmatile_39',['MMATile',['../structmlx_1_1steel_1_1_m_m_a_tile.html#aa3fb310dd08ec23c334511f7b316d1b6',1,'mlx::steel::MMATile']]], - ['move_5fshared_5fbuffer_40',['move_shared_buffer',['../classmlx_1_1core_1_1array.html#acce00db63e0f3d80f797b02397ade836',1,'mlx::core::array::move_shared_buffer(array other, const std::vector< size_t > &strides, Flags flags, size_t data_size, size_t offset=0)'],['../classmlx_1_1core_1_1array.html#a38d7ad605f8282e5e49d0c09e0555c78',1,'mlx::core::array::move_shared_buffer(array other)']]], - ['moveaxis_41',['moveaxis',['../group__ops.html#ga24067d10a842db2c9d509ea48135a2c3',1,'mlx::core']]], - ['mpinplace_42',['MPINPLACE',['../namespacepocketfft_1_1detail.html#af5eedf3cdfc83c0a30807092c39a9ce2',1,'pocketfft::detail']]], - ['mtl_5fdevice_43',['mtl_device',['../classmlx_1_1core_1_1metal_1_1_device.html#a31dba377f2be44a746db10d1b9367653',1,'mlx::core::metal::Device']]], - ['mtl_5fresidency_5fset_44',['mtl_residency_set',['../classmlx_1_1core_1_1metal_1_1_residency_set.html#ac4bfe5ef5e2eaebc458a1ed1953d15e9',1,'mlx::core::metal::ResidencySet']]], - ['multi_5fiter_45',['multi_iter',['../classpocketfft_1_1detail_1_1multi__iter.html#a9be43bb18840202da6d17988fccc64b9',1,'pocketfft::detail::multi_iter']]], - ['multiply_46',['Multiply',['../classmlx_1_1core_1_1_multiply.html#aca5c50f900321f3eb4d6fbcbc225c00c',1,'mlx::core::Multiply']]], - ['multiply_47',['multiply',['../group__ops.html#gaf57392e641640b5d06e4c99518391c38',1,'mlx::core']]], - ['multivariate_5fnormal_48',['multivariate_normal',['../namespacemlx_1_1core_1_1random.html#a8c37da3c1c0c561cad7499d6d9db81fb',1,'mlx::core::random']]] + ['maybeinsertbarrier_14',['maybeInsertBarrier',['../structmlx_1_1core_1_1metal_1_1_command_encoder.html#ad538ae88f90560063f9ba502e2795991',1,'mlx::core::metal::CommandEncoder']]], + ['mb_5fblock_5fmerge_15',['mb_block_merge',['../sort_8h.html#ab381cd57f344bc7304ab580bfdc78807',1,'sort.h']]], + ['mb_5fblock_5fpartition_16',['mb_block_partition',['../sort_8h.html#a32cbe4163b8b0f5cb2c97b256119a4b2',1,'sort.h']]], + ['mb_5fblock_5fsort_17',['mb_block_sort',['../sort_8h.html#aa48ff1aff1e9dc1301b6781aa0721d6b',1,'sort.h']]], + ['mean_18',['mean',['../group__ops.html#gade46e768fd46b8b640eb16f26abeecef',1,'mlx::core::mean(const array &a, bool keepdims, StreamOrDevice s={})'],['../group__ops.html#ga52b59fdd8e8430538e564f5bbcfa31e6',1,'mlx::core::mean(const array &a, StreamOrDevice s={})'],['../group__ops.html#ga066161f3d3e395a1d76c638cb680d444',1,'mlx::core::mean(const array &a, const std::vector< int > &axes, bool keepdims=false, StreamOrDevice s={})'],['../group__ops.html#ga45fba73eab0e3b6e128ed3ce2f43a5da',1,'mlx::core::mean(const array &a, int axis, bool keepdims=false, StreamOrDevice s={})']]], + ['median3_19',['median3',['../namespacemetal.html#aa3ff49457ce3c93fc1c0897fd1525157',1,'metal::median3()'],['../namespacemetal_1_1fast.html#a742b55f1e4369921ee7f60d70185bfbc',1,'metal::fast::median3()'],['../namespacemetal_1_1precise.html#a14555ff99c4388493fec48e070144ae2',1,'metal::precise::median3()']]], + ['merge_5fpartition_20',['merge_partition',['../struct_block_merge_sort.html#ab2300cbecb23f3433bad888924c831ca',1,'BlockMergeSort::merge_partition()'],['../struct_kernel_multi_block_merge_sort.html#ab15895b4233aba0e279cc44a07a201fe',1,'KernelMultiBlockMergeSort::merge_partition()']]], + ['merge_5fstep_21',['merge_step',['../struct_block_merge_sort.html#ab65f190edf1851b37c39ad49ce99a43c',1,'BlockMergeSort']]], + ['meshgrid_22',['meshgrid',['../group__ops.html#ga577c911618575314de63d1060656a26e',1,'mlx::core']]], + ['metal_5fkernel_23',['metal_kernel',['../namespacemlx_1_1core_1_1fast.html#ab16436b465dc10ce472193d541d8426e',1,'mlx::core::fast']]], + ['min_24',['min',['../structmetal_1_1__numeric__limits__impl_3_01bfloat16__t_01_4.html#adaed80031f5ca0ff69d30ec4c5d0c98f',1,'metal::_numeric_limits_impl< bfloat16_t >::min()'],['../namespacemetal.html#a6653b28c9473087141eddce39878d4d3',1,'metal::min()'],['../namespacemetal_1_1fast.html#a3e958e56a4712687c381a0b64d123e61',1,'metal::fast::min()'],['../namespacemetal_1_1precise.html#afed0da2f7df3505b5dffa2389c3cb36e',1,'metal::precise::min()'],['../group__ops.html#gab27599802617a4c8f9964ab5f4ffee12',1,'mlx::core::min(const array &a, bool keepdims, StreamOrDevice s={})'],['../group__ops.html#ga0140b91e9cdfc3fef0da8e332f65a9e8',1,'mlx::core::min(const array &a, StreamOrDevice s={})'],['../group__ops.html#ga6efb83cd46436678c8f8c4af15cc00f5',1,'mlx::core::min(const array &a, const std::vector< int > &axes, bool keepdims=false, StreamOrDevice s={})'],['../group__ops.html#ga36fa315eef677f4143868f552cd26d03',1,'mlx::core::min(const array &a, int axis, bool keepdims=false, StreamOrDevice s={})']]], + ['min3_25',['min3',['../namespacemetal.html#a005510c8c0f964ce2b8aad3ba76a7a3f',1,'metal::min3()'],['../namespacemetal_1_1fast.html#a606a4c1b34ce05ea89ca5af81724036f',1,'metal::fast::min3()'],['../namespacemetal_1_1precise.html#a4d37ce31c3549ca4772a4ee29798e231',1,'metal::precise::min3()']]], + ['minimum_26',['Minimum',['../classmlx_1_1core_1_1_minimum.html#ab0f2ce17108df44b82cff68886b0f6f5',1,'mlx::core::Minimum']]], + ['minimum_27',['minimum',['../group__ops.html#ga49ba00c090f81f331c91b0c97040bce0',1,'mlx::core']]], + ['mlx_5fatomic_5fcompare_5fexchange_5fweak_5fexplicit_28',['mlx_atomic_compare_exchange_weak_explicit',['../atomic_8h.html#ad7f32327ff66354cfa2f0cfdac79316f',1,'mlx_atomic_compare_exchange_weak_explicit(device mlx_atomic< T > *object, thread T *expected, T val, size_t offset): atomic.h'],['../atomic_8h.html#aa8f47b2e9b95d4b00ad51f08b070deb5',1,'mlx_atomic_compare_exchange_weak_explicit(device mlx_atomic< T > *object, thread uint *expected, uint val, size_t offset): atomic.h']]], + ['mlx_5fatomic_5ffetch_5fadd_5fexplicit_29',['mlx_atomic_fetch_add_explicit',['../atomic_8h.html#aad448d9e06e001700b65ca8317216a3b',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fand_5fexplicit_30',['mlx_atomic_fetch_and_explicit',['../atomic_8h.html#a253e3c870c0ddc7c28ab2f6ca2c3eae5',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fmax_5fexplicit_31',['mlx_atomic_fetch_max_explicit',['../atomic_8h.html#ac480f2b459a8ad9095cee353e152d00c',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fmax_5fexplicit_3c_20float_20_3e_32',['mlx_atomic_fetch_max_explicit< float >',['../atomic_8h.html#a1dce2abfa16417122c4d2bf261129ae4',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fmin_5fexplicit_33',['mlx_atomic_fetch_min_explicit',['../atomic_8h.html#a2ec33dca0039bd944d73d1c2b378cc19',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fmin_5fexplicit_3c_20float_20_3e_34',['mlx_atomic_fetch_min_explicit< float >',['../atomic_8h.html#ab7d1dc49f319f239b7ee0b7c72976dd0',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5fmul_5fexplicit_35',['mlx_atomic_fetch_mul_explicit',['../atomic_8h.html#adfdbea60436f14f1af9ce36e2a0a77a3',1,'atomic.h']]], + ['mlx_5fatomic_5ffetch_5for_5fexplicit_36',['mlx_atomic_fetch_or_explicit',['../atomic_8h.html#ab7391f197001471e4788312bdb6ab37a',1,'atomic.h']]], + ['mlx_5fatomic_5fload_5fexplicit_37',['mlx_atomic_load_explicit',['../atomic_8h.html#a253a4e8c2c5768a069e2791b627dfc99',1,'atomic.h']]], + ['mlx_5fatomic_5fstore_5fexplicit_38',['mlx_atomic_store_explicit',['../atomic_8h.html#a0ae453140b0819a4c02f265334de98c0',1,'atomic.h']]], + ['mma_39',['mma',['../structmlx_1_1steel_1_1_base_m_m_a_frag_3_01_t_00_018_00_018_01_4.html#a8028512f5a3d2b6acaf966be529627a3',1,'mlx::steel::BaseMMAFrag< T, 8, 8 >::mma(thread frag_type &D, thread frag_type &A, thread frag_type &B, thread frag_type &C)'],['../structmlx_1_1steel_1_1_base_m_m_a_frag_3_01_t_00_018_00_018_01_4.html#a1868f57d57c8adedab2c58492ec76946',1,'mlx::steel::BaseMMAFrag< T, 8, 8 >::mma(thread mat_type &D, thread mat_type &A, thread mat_type &B, thread mat_type &C)'],['../structmlx_1_1steel_1_1_block_m_m_a.html#a6a2c2a6d5e767d52c41b42a9d36086b0',1,'mlx::steel::BlockMMA::mma()']]], + ['mmatile_40',['MMATile',['../structmlx_1_1steel_1_1_m_m_a_tile.html#aa3fb310dd08ec23c334511f7b316d1b6',1,'mlx::steel::MMATile']]], + ['move_5fshared_5fbuffer_41',['move_shared_buffer',['../classmlx_1_1core_1_1array.html#acce00db63e0f3d80f797b02397ade836',1,'mlx::core::array::move_shared_buffer(array other, const std::vector< size_t > &strides, Flags flags, size_t data_size, size_t offset=0)'],['../classmlx_1_1core_1_1array.html#a38d7ad605f8282e5e49d0c09e0555c78',1,'mlx::core::array::move_shared_buffer(array other)']]], + ['moveaxis_42',['moveaxis',['../group__ops.html#ga24067d10a842db2c9d509ea48135a2c3',1,'mlx::core']]], + ['mpinplace_43',['MPINPLACE',['../namespacepocketfft_1_1detail.html#af5eedf3cdfc83c0a30807092c39a9ce2',1,'pocketfft::detail']]], + ['mtl_5fdevice_44',['mtl_device',['../classmlx_1_1core_1_1metal_1_1_device.html#a31dba377f2be44a746db10d1b9367653',1,'mlx::core::metal::Device']]], + ['mtl_5fresidency_5fset_45',['mtl_residency_set',['../classmlx_1_1core_1_1metal_1_1_residency_set.html#ac4bfe5ef5e2eaebc458a1ed1953d15e9',1,'mlx::core::metal::ResidencySet']]], + ['multi_5fiter_46',['multi_iter',['../classpocketfft_1_1detail_1_1multi__iter.html#a9be43bb18840202da6d17988fccc64b9',1,'pocketfft::detail::multi_iter']]], + ['multiply_47',['Multiply',['../classmlx_1_1core_1_1_multiply.html#aca5c50f900321f3eb4d6fbcbc225c00c',1,'mlx::core::Multiply']]], + ['multiply_48',['multiply',['../group__ops.html#gaf57392e641640b5d06e4c99518391c38',1,'mlx::core']]], + ['multivariate_5fnormal_49',['multivariate_normal',['../namespacemlx_1_1core_1_1random.html#a8c37da3c1c0c561cad7499d6d9db81fb',1,'mlx::core::random']]] ]; diff --git a/docs/build/html/searchindex.js b/docs/build/html/searchindex.js index 7e7aaf386..09019b7d5 100644 --- a/docs/build/html/searchindex.js +++ b/docs/build/html/searchindex.js @@ -1 +1 @@ -Search.setIndex({"alltitles": {"A Simple Example": [[487, "a-simple-example"]], "Array": [[316, null]], "Attention layer": [[5, "attention-layer"]], "Automatic Differentiation": [[481, "automatic-differentiation"]], "Automatic Vectorization": [[481, "automatic-vectorization"]], "Basics": [[485, "basics"]], "Basics of Compile": [[479, "basics-of-compile"]], "Binary Size Minimization": [[8, "binary-size-minimization"]], "Binding to Python": [[2, "binding-to-python"]], "Build Options": [[8, "id3"]], "Build Requirements": [[8, "build-requirements"]], "Build and Install": [[8, null]], "Build from source": [[8, "build-from-source"]], "Building and Binding": [[2, "building-and-binding"]], "Building with CMake": [[2, "building-with-cmake"]], "Building with setuptools": [[2, "building-with-setuptools"]], "C++ API": [[8, "c-api"]], "C++ API Reference": [[7, null]], "Common Optimizers": [[473, null]], "Compilation": [[479, null]], "Compiling Training Graphs": [[479, "compiling-training-graphs"]], "Complex Example": [[1, "complex-example"]], "Conversion to NumPy and Other Frameworks": [[484, null]], "Converting the weights": [[5, "converting-the-weights"]], "Custom Extensions in MLX": [[2, null]], "Custom Metal Kernels": [[1, null]], "Data Types": [[317, null]], "Debugging": [[479, "debugging"]], "Devices and Streams": [[318, null]], "Differences from NumPy": [[482, "differences-from-numpy"]], "Distributed Communication": [[319, null], [480, null]], "Download the code": [[2, null], [5, null]], "Encoder layer": [[5, "encoder-layer"]], "Example Speedup": [[479, "example-speedup"]], "Examples": [[7, null]], "FFT": [[321, null]], "Fast": [[320, null]], "Full model": [[5, "full-model"]], "Function Transforms": [[481, null]], "Function and Graph Transformations": [[485, "function-and-graph-transformations"]], "Functions": [[448, null]], "Further Reading": [[7, null]], "Generation": [[5, "generation"]], "Getting Started": [[480, "getting-started"]], "Grid Sample VJP": [[1, "grid-sample-vjp"]], "Implementing the CPU Back-end": [[2, "implementing-the-cpu-back-end"]], "Implementing the GPU Back-end": [[2, "implementing-the-gpu-back-end"]], "Implementing the Primitive": [[2, "implementing-the-primitive"]], "Implementing the model": [[5, "implementing-the-model"]], "In Place Updates": [[482, "in-place-updates"]], "Indexing Arrays": [[482, null]], "Initializers": [[449, null]], "Inspecting Modules": [[324, "inspecting-modules"]], "Install": [[7, null]], "Installing MPI": [[480, "installing-mpi"]], "Introducing the Example": [[2, "introducing-the-example"]], "JAX": [[484, "jax"]], "LLM inference": [[5, null]], "Layers": [[450, null]], "Lazy Evaluation": [[483, null]], "Linear Algebra": [[322, null]], "Linear Regression": [[4, null]], "Loss Functions": [[451, null]], "MLX": [[7, null]], "Metal": [[323, null]], "Metal Debugger": [[3, null]], "Metal not found": [[8, "metal-not-found"]], "Module": [[452, null]], "Multi-Layer Perceptron": [[6, null]], "Neural Networks": [[324, null]], "Only Compute What You Use": [[483, "only-compute-what-you-use"]], "Operations": [[0, null], [2, "operations"], [453, null]], "Operations and Primitives": [[2, "operations-and-primitives"]], "Optimizer": [[474, null]], "Optimizers": [[454, null]], "Parameters": [[324, "parameters"]], "Primitive Transforms": [[2, "primitive-transforms"]], "Primitives": [[2, "primitives"]], "Pure Functions": [[479, "pure-functions"]], "Putting it all together": [[5, "putting-it-all-together"]], "PyTorch": [[484, "pytorch"]], "Python API": [[8, "python-api"]], "Python API Reference": [[7, null]], "Python Installation": [[8, "python-installation"]], "Quick Start Guide": [[485, null]], "Quick Start with Neural Networks": [[324, "quick-start-with-neural-networks"]], "Random": [[476, null]], "Results": [[2, "results"]], "Saving and Loading": [[454, "saving-and-loading"]], "Saving and Loading Arrays": [[486, null]], "Schedulers": [[475, null]], "Scripts": [[2, "scripts"], [5, "scripts"]], "Serialization Formats": [[486, "id1"]], "Setting up Remote Hosts": [[480, "setting-up-remote-hosts"]], "Simple Example": [[1, "simple-example"]], "Specifying the Stream": [[488, "specifying-the-stream"]], "Supported Data Types": [[317, "id2"]], "TensorFlow": [[484, "tensorflow"]], "The Module Class": [[324, "the-module-class"]], "Training Example": [[480, "training-example"]], "Transformations with Compile": [[479, "transformations-with-compile"]], "Transforming Compute Graphs": [[483, "transforming-compute-graphs"]], "Transforms": [[477, null]], "Tree Utils": [[478, null]], "Troubleshooting": [[8, "troubleshooting"], [8, "id2"]], "Tuning All Reduce": [[480, "tuning-all-reduce"]], "Unified Memory": [[487, null]], "Updating the Parameters": [[324, "updating-the-parameters"]], "Usage": [[2, "usage"], [7, null]], "Using Shape/Strides": [[1, "using-shape-strides"]], "Using Streams": [[488, null]], "Using the Primitive": [[2, "using-the-primitive"]], "Value and Grad": [[324, "value-and-grad"]], "Weight loading and benchmarking": [[5, "weight-loading-and-benchmarking"]], "When to Evaluate": [[483, "when-to-evaluate"]], "Why Lazy Evaluation": [[483, "why-lazy-evaluation"]], "Xcode Workflow": [[3, "xcode-workflow"]], "mlx.core.Device": [[9, null]], "mlx.core.Dtype": [[10, null]], "mlx.core.DtypeCategory": [[11, null]], "mlx.core.Stream": [[315, null]], "mlx.core.abs": [[12, null]], "mlx.core.add": [[13, null]], "mlx.core.addmm": [[14, null]], "mlx.core.all": [[15, null]], "mlx.core.allclose": [[16, null]], "mlx.core.any": [[17, null]], "mlx.core.arange": [[18, null]], "mlx.core.arccos": [[19, null]], "mlx.core.arccosh": [[20, null]], "mlx.core.arcsin": [[21, null]], "mlx.core.arcsinh": [[22, null]], "mlx.core.arctan": [[23, null]], "mlx.core.arctan2": [[24, null]], "mlx.core.arctanh": [[25, null]], "mlx.core.argmax": [[26, null]], "mlx.core.argmin": [[27, null]], "mlx.core.argpartition": [[28, null]], "mlx.core.argsort": [[29, null]], "mlx.core.array": [[30, null]], "mlx.core.array.T": [[31, null]], "mlx.core.array.abs": [[32, null]], "mlx.core.array.all": [[33, null]], "mlx.core.array.any": [[34, null]], "mlx.core.array.argmax": [[35, null]], "mlx.core.array.argmin": [[36, null]], "mlx.core.array.astype": [[37, null]], "mlx.core.array.at": [[38, null]], "mlx.core.array.conj": [[39, null]], "mlx.core.array.cos": [[40, null]], "mlx.core.array.cummax": [[41, null]], "mlx.core.array.cummin": [[42, null]], "mlx.core.array.cumprod": [[43, null]], "mlx.core.array.cumsum": [[44, null]], "mlx.core.array.diag": [[45, null]], "mlx.core.array.diagonal": [[46, null]], "mlx.core.array.dtype": [[47, null]], "mlx.core.array.exp": [[48, null]], "mlx.core.array.flatten": [[49, null]], "mlx.core.array.item": [[50, null]], "mlx.core.array.itemsize": [[51, null]], "mlx.core.array.log": [[52, null]], "mlx.core.array.log10": [[53, null]], "mlx.core.array.log1p": [[54, null]], "mlx.core.array.log2": [[55, null]], "mlx.core.array.logsumexp": [[56, null]], "mlx.core.array.max": [[57, null]], "mlx.core.array.mean": [[58, null]], "mlx.core.array.min": [[59, null]], "mlx.core.array.moveaxis": [[60, null]], "mlx.core.array.nbytes": [[61, null]], "mlx.core.array.ndim": [[62, null]], "mlx.core.array.prod": [[63, null]], "mlx.core.array.reciprocal": [[64, null]], "mlx.core.array.reshape": [[65, null]], "mlx.core.array.round": [[66, null]], "mlx.core.array.rsqrt": [[67, null]], "mlx.core.array.shape": [[68, null]], "mlx.core.array.sin": [[69, null]], "mlx.core.array.size": [[70, null]], "mlx.core.array.split": [[71, null]], "mlx.core.array.sqrt": [[72, null]], "mlx.core.array.square": [[73, null]], "mlx.core.array.squeeze": [[74, null]], "mlx.core.array.std": [[75, null]], "mlx.core.array.sum": [[76, null]], "mlx.core.array.swapaxes": [[77, null]], "mlx.core.array.tolist": [[78, null]], "mlx.core.array.transpose": [[79, null]], "mlx.core.array.var": [[80, null]], "mlx.core.array.view": [[81, null]], "mlx.core.array_equal": [[82, null]], "mlx.core.as_strided": [[83, null]], "mlx.core.atleast_1d": [[84, null]], "mlx.core.atleast_2d": [[85, null]], "mlx.core.atleast_3d": [[86, null]], "mlx.core.bitwise_and": [[87, null]], "mlx.core.bitwise_or": [[88, null]], "mlx.core.bitwise_xor": [[89, null]], "mlx.core.block_masked_mm": [[90, null]], "mlx.core.broadcast_to": [[91, null]], "mlx.core.ceil": [[92, null]], "mlx.core.clip": [[93, null]], "mlx.core.compile": [[94, null]], "mlx.core.concatenate": [[95, null]], "mlx.core.conj": [[96, null]], "mlx.core.conjugate": [[97, null]], "mlx.core.conv1d": [[98, null]], "mlx.core.conv2d": [[99, null]], "mlx.core.conv3d": [[100, null]], "mlx.core.conv_general": [[101, null]], "mlx.core.conv_transpose1d": [[102, null]], "mlx.core.conv_transpose2d": [[103, null]], "mlx.core.conv_transpose3d": [[104, null]], "mlx.core.convolve": [[105, null]], "mlx.core.cos": [[106, null]], "mlx.core.cosh": [[107, null]], "mlx.core.cummax": [[108, null]], "mlx.core.cummin": [[109, null]], "mlx.core.cumprod": [[110, null]], "mlx.core.cumsum": [[111, null]], "mlx.core.custom_function": [[112, null]], "mlx.core.default_device": [[113, null]], "mlx.core.default_stream": [[114, null]], "mlx.core.degrees": [[115, null]], "mlx.core.dequantize": [[116, null]], "mlx.core.diag": [[117, null]], "mlx.core.diagonal": [[118, null]], "mlx.core.disable_compile": [[119, null]], "mlx.core.distributed.Group": [[120, null]], "mlx.core.distributed.all_gather": [[121, null]], "mlx.core.distributed.all_sum": [[122, null]], "mlx.core.distributed.init": [[123, null]], "mlx.core.distributed.is_available": [[124, null]], "mlx.core.distributed.recv": [[125, null]], "mlx.core.distributed.recv_like": [[126, null]], "mlx.core.distributed.send": [[127, null]], "mlx.core.divide": [[128, null]], "mlx.core.divmod": [[129, null]], "mlx.core.einsum": [[130, null]], "mlx.core.einsum_path": [[131, null]], "mlx.core.enable_compile": [[132, null]], "mlx.core.equal": [[133, null]], "mlx.core.erf": [[134, null]], "mlx.core.erfinv": [[135, null]], "mlx.core.eval": [[136, null]], "mlx.core.exp": [[137, null]], "mlx.core.expand_dims": [[138, null]], "mlx.core.expm1": [[139, null]], "mlx.core.eye": [[140, null]], "mlx.core.fast.affine_quantize": [[141, null]], "mlx.core.fast.layer_norm": [[142, null]], "mlx.core.fast.metal_kernel": [[143, null]], "mlx.core.fast.rms_norm": [[144, null]], "mlx.core.fast.rope": [[145, null]], "mlx.core.fast.scaled_dot_product_attention": [[146, null]], "mlx.core.fft.fft": [[147, null]], "mlx.core.fft.fft2": [[148, null]], "mlx.core.fft.fftn": [[149, null]], "mlx.core.fft.ifft": [[150, null]], "mlx.core.fft.ifft2": [[151, null]], "mlx.core.fft.ifftn": [[152, null]], "mlx.core.fft.irfft": [[153, null]], "mlx.core.fft.irfft2": [[154, null]], "mlx.core.fft.irfftn": [[155, null]], "mlx.core.fft.rfft": [[156, null]], "mlx.core.fft.rfft2": [[157, null]], "mlx.core.fft.rfftn": [[158, null]], "mlx.core.flatten": [[159, null]], "mlx.core.floor": [[160, null]], "mlx.core.floor_divide": [[161, null]], "mlx.core.full": [[162, null]], "mlx.core.gather_mm": [[163, null]], "mlx.core.gather_qmm": [[164, null]], "mlx.core.grad": [[165, null]], "mlx.core.greater": [[166, null]], "mlx.core.greater_equal": [[167, null]], "mlx.core.hadamard_transform": [[168, null]], "mlx.core.identity": [[169, null]], "mlx.core.imag": [[170, null]], "mlx.core.inner": [[171, null]], "mlx.core.isclose": [[172, null]], "mlx.core.isfinite": [[173, null]], "mlx.core.isinf": [[174, null]], "mlx.core.isnan": [[175, null]], "mlx.core.isneginf": [[176, null]], "mlx.core.isposinf": [[177, null]], "mlx.core.issubdtype": [[178, null]], "mlx.core.jvp": [[179, null]], "mlx.core.left_shift": [[180, null]], "mlx.core.less": [[181, null]], "mlx.core.less_equal": [[182, null]], "mlx.core.linalg.cholesky": [[183, null]], "mlx.core.linalg.cholesky_inv": [[184, null]], "mlx.core.linalg.cross": [[185, null]], "mlx.core.linalg.eigh": [[186, null]], "mlx.core.linalg.eigvalsh": [[187, null]], "mlx.core.linalg.inv": [[188, null]], "mlx.core.linalg.norm": [[189, null]], "mlx.core.linalg.qr": [[190, null]], "mlx.core.linalg.svd": [[191, null]], "mlx.core.linalg.tri_inv": [[192, null]], "mlx.core.linspace": [[193, null]], "mlx.core.load": [[194, null]], "mlx.core.log": [[195, null]], "mlx.core.log10": [[196, null]], "mlx.core.log1p": [[197, null]], "mlx.core.log2": [[198, null]], "mlx.core.logaddexp": [[199, null]], "mlx.core.logical_and": [[200, null]], "mlx.core.logical_not": [[201, null]], "mlx.core.logical_or": [[202, null]], "mlx.core.logsumexp": [[203, null]], "mlx.core.matmul": [[204, null]], "mlx.core.max": [[205, null]], "mlx.core.maximum": [[206, null]], "mlx.core.mean": [[207, null]], "mlx.core.meshgrid": [[208, null]], "mlx.core.metal.clear_cache": [[209, null]], "mlx.core.metal.device_info": [[210, null]], "mlx.core.metal.get_active_memory": [[211, null]], "mlx.core.metal.get_cache_memory": [[212, null]], "mlx.core.metal.get_peak_memory": [[213, null]], "mlx.core.metal.is_available": [[214, null]], "mlx.core.metal.reset_peak_memory": [[215, null]], "mlx.core.metal.set_cache_limit": [[216, null]], "mlx.core.metal.set_memory_limit": [[217, null]], "mlx.core.metal.set_wired_limit": [[218, null]], "mlx.core.metal.start_capture": [[219, null]], "mlx.core.metal.stop_capture": [[220, null]], "mlx.core.min": [[221, null]], "mlx.core.minimum": [[222, null]], "mlx.core.moveaxis": [[223, null]], "mlx.core.multiply": [[224, null]], "mlx.core.nan_to_num": [[225, null]], "mlx.core.negative": [[226, null]], "mlx.core.new_stream": [[227, null]], "mlx.core.not_equal": [[228, null]], "mlx.core.ones": [[229, null]], "mlx.core.ones_like": [[230, null]], "mlx.core.outer": [[231, null]], "mlx.core.pad": [[232, null]], "mlx.core.partition": [[233, null]], "mlx.core.power": [[234, null]], "mlx.core.prod": [[235, null]], "mlx.core.put_along_axis": [[236, null]], "mlx.core.quantize": [[237, null]], "mlx.core.quantized_matmul": [[238, null]], "mlx.core.radians": [[239, null]], "mlx.core.random.bernoulli": [[240, null]], "mlx.core.random.categorical": [[241, null]], "mlx.core.random.gumbel": [[242, null]], "mlx.core.random.key": [[243, null]], "mlx.core.random.laplace": [[244, null]], "mlx.core.random.multivariate_normal": [[245, null]], "mlx.core.random.normal": [[246, null]], "mlx.core.random.permutation": [[247, null]], "mlx.core.random.randint": [[248, null]], "mlx.core.random.seed": [[249, null]], "mlx.core.random.split": [[250, null]], "mlx.core.random.truncated_normal": [[251, null]], "mlx.core.random.uniform": [[252, null]], "mlx.core.real": [[253, null]], "mlx.core.reciprocal": [[254, null]], "mlx.core.remainder": [[255, null]], "mlx.core.repeat": [[256, null]], "mlx.core.reshape": [[257, null]], "mlx.core.right_shift": [[258, null]], "mlx.core.roll": [[259, null]], "mlx.core.round": [[260, null]], "mlx.core.rsqrt": [[261, null]], "mlx.core.save": [[262, null]], "mlx.core.save_gguf": [[263, null]], "mlx.core.save_safetensors": [[264, null]], "mlx.core.savez": [[265, null]], "mlx.core.savez_compressed": [[266, null]], "mlx.core.set_default_device": [[267, null]], "mlx.core.set_default_stream": [[268, null]], "mlx.core.sigmoid": [[269, null]], "mlx.core.sign": [[270, null]], "mlx.core.sin": [[271, null]], "mlx.core.sinh": [[272, null]], "mlx.core.softmax": [[273, null]], "mlx.core.sort": [[274, null]], "mlx.core.split": [[275, null]], "mlx.core.sqrt": [[276, null]], "mlx.core.square": [[277, null]], "mlx.core.squeeze": [[278, null]], "mlx.core.stack": [[279, null]], "mlx.core.std": [[280, null]], "mlx.core.stop_gradient": [[281, null]], "mlx.core.stream": [[282, null]], "mlx.core.subtract": [[283, null]], "mlx.core.sum": [[284, null]], "mlx.core.swapaxes": [[285, null]], "mlx.core.synchronize": [[286, null]], "mlx.core.take": [[287, null]], "mlx.core.take_along_axis": [[288, null]], "mlx.core.tan": [[289, null]], "mlx.core.tanh": [[290, null]], "mlx.core.tensordot": [[291, null]], "mlx.core.tile": [[292, null]], "mlx.core.topk": [[293, null]], "mlx.core.trace": [[294, null]], "mlx.core.transpose": [[295, null]], "mlx.core.tri": [[296, null]], "mlx.core.tril": [[297, null]], "mlx.core.triu": [[298, null]], "mlx.core.value_and_grad": [[299, null]], "mlx.core.var": [[300, null]], "mlx.core.view": [[301, null]], "mlx.core.vjp": [[302, null]], "mlx.core.vmap": [[303, null]], "mlx.core.where": [[304, null]], "mlx.core.zeros": [[305, null]], "mlx.core.zeros_like": [[306, null]], "mlx.nn.ALiBi": [[325, null]], "mlx.nn.AvgPool1d": [[326, null]], "mlx.nn.AvgPool2d": [[327, null]], "mlx.nn.BatchNorm": [[328, null]], "mlx.nn.CELU": [[329, null]], "mlx.nn.Conv1d": [[330, null]], "mlx.nn.Conv2d": [[331, null]], "mlx.nn.Conv3d": [[332, null]], "mlx.nn.ConvTranspose1d": [[333, null]], "mlx.nn.ConvTranspose2d": [[334, null]], "mlx.nn.ConvTranspose3d": [[335, null]], "mlx.nn.Dropout": [[336, null]], "mlx.nn.Dropout2d": [[337, null]], "mlx.nn.Dropout3d": [[338, null]], "mlx.nn.ELU": [[339, null]], "mlx.nn.Embedding": [[340, null]], "mlx.nn.GELU": [[341, null]], "mlx.nn.GLU": [[342, null]], "mlx.nn.GRU": [[343, null]], "mlx.nn.GroupNorm": [[344, null]], "mlx.nn.HardShrink": [[345, null]], "mlx.nn.HardTanh": [[346, null]], "mlx.nn.Hardswish": [[347, null]], "mlx.nn.InstanceNorm": [[348, null]], "mlx.nn.LSTM": [[349, null]], "mlx.nn.LayerNorm": [[350, null]], "mlx.nn.LeakyReLU": [[351, null]], "mlx.nn.Linear": [[352, null]], "mlx.nn.LogSigmoid": [[353, null]], "mlx.nn.LogSoftmax": [[354, null]], "mlx.nn.MaxPool1d": [[355, null]], "mlx.nn.MaxPool2d": [[356, null]], "mlx.nn.Mish": [[357, null]], "mlx.nn.Module.apply": [[358, null]], "mlx.nn.Module.apply_to_modules": [[359, null]], "mlx.nn.Module.children": [[360, null]], "mlx.nn.Module.eval": [[361, null]], "mlx.nn.Module.filter_and_map": [[362, null]], "mlx.nn.Module.freeze": [[363, null]], "mlx.nn.Module.leaf_modules": [[364, null]], "mlx.nn.Module.load_weights": [[365, null]], "mlx.nn.Module.modules": [[366, null]], "mlx.nn.Module.named_modules": [[367, null]], "mlx.nn.Module.parameters": [[368, null]], "mlx.nn.Module.save_weights": [[369, null]], "mlx.nn.Module.set_dtype": [[370, null]], "mlx.nn.Module.state": [[371, null]], "mlx.nn.Module.train": [[372, null]], "mlx.nn.Module.trainable_parameters": [[373, null]], "mlx.nn.Module.training": [[374, null]], "mlx.nn.Module.unfreeze": [[375, null]], "mlx.nn.Module.update": [[376, null]], "mlx.nn.Module.update_modules": [[377, null]], "mlx.nn.MultiHeadAttention": [[378, null]], "mlx.nn.PReLU": [[379, null]], "mlx.nn.QuantizedEmbedding": [[380, null]], "mlx.nn.QuantizedLinear": [[381, null]], "mlx.nn.RMSNorm": [[382, null]], "mlx.nn.RNN": [[383, null]], "mlx.nn.ReLU": [[384, null]], "mlx.nn.ReLU6": [[385, null]], "mlx.nn.RoPE": [[386, null]], "mlx.nn.SELU": [[387, null]], "mlx.nn.Sequential": [[388, null]], "mlx.nn.SiLU": [[389, null]], "mlx.nn.Sigmoid": [[390, null]], "mlx.nn.SinusoidalPositionalEncoding": [[391, null]], "mlx.nn.Softmax": [[392, null]], "mlx.nn.Softmin": [[393, null]], "mlx.nn.Softplus": [[394, null]], "mlx.nn.Softshrink": [[395, null]], "mlx.nn.Softsign": [[396, null]], "mlx.nn.Step": [[397, null]], "mlx.nn.Tanh": [[398, null]], "mlx.nn.Transformer": [[399, null]], "mlx.nn.Upsample": [[400, null]], "mlx.nn.celu": [[409, null]], "mlx.nn.elu": [[410, null]], "mlx.nn.gelu": [[411, null]], "mlx.nn.gelu_approx": [[412, null]], "mlx.nn.gelu_fast_approx": [[413, null]], "mlx.nn.glu": [[414, null]], "mlx.nn.hard_shrink": [[415, null]], "mlx.nn.hard_tanh": [[416, null]], "mlx.nn.hardswish": [[417, null]], "mlx.nn.init.constant": [[401, null]], "mlx.nn.init.glorot_normal": [[402, null]], "mlx.nn.init.glorot_uniform": [[403, null]], "mlx.nn.init.he_normal": [[404, null]], "mlx.nn.init.he_uniform": [[405, null]], "mlx.nn.init.identity": [[406, null]], "mlx.nn.init.normal": [[407, null]], "mlx.nn.init.uniform": [[408, null]], "mlx.nn.leaky_relu": [[418, null]], "mlx.nn.log_sigmoid": [[419, null]], "mlx.nn.log_softmax": [[420, null]], "mlx.nn.losses.binary_cross_entropy": [[421, null]], "mlx.nn.losses.cosine_similarity_loss": [[422, null]], "mlx.nn.losses.cross_entropy": [[423, null]], "mlx.nn.losses.gaussian_nll_loss": [[424, null]], "mlx.nn.losses.hinge_loss": [[425, null]], "mlx.nn.losses.huber_loss": [[426, null]], "mlx.nn.losses.kl_div_loss": [[427, null]], "mlx.nn.losses.l1_loss": [[428, null]], "mlx.nn.losses.log_cosh_loss": [[429, null]], "mlx.nn.losses.margin_ranking_loss": [[430, null]], "mlx.nn.losses.mse_loss": [[431, null]], "mlx.nn.losses.nll_loss": [[432, null]], "mlx.nn.losses.smooth_l1_loss": [[433, null]], "mlx.nn.losses.triplet_loss": [[434, null]], "mlx.nn.mish": [[435, null]], "mlx.nn.prelu": [[436, null]], "mlx.nn.quantize": [[307, null]], "mlx.nn.relu": [[437, null]], "mlx.nn.relu6": [[438, null]], "mlx.nn.selu": [[439, null]], "mlx.nn.sigmoid": [[440, null]], "mlx.nn.silu": [[441, null]], "mlx.nn.softmax": [[442, null]], "mlx.nn.softmin": [[443, null]], "mlx.nn.softplus": [[444, null]], "mlx.nn.softshrink": [[445, null]], "mlx.nn.step": [[446, null]], "mlx.nn.tanh": [[447, null]], "mlx.nn.value_and_grad": [[308, null]], "mlx.optimizers.AdaDelta": [[455, null]], "mlx.optimizers.Adafactor": [[456, null]], "mlx.optimizers.Adagrad": [[457, null]], "mlx.optimizers.Adam": [[458, null]], "mlx.optimizers.AdamW": [[459, null]], "mlx.optimizers.Adamax": [[460, null]], "mlx.optimizers.Lion": [[461, null]], "mlx.optimizers.Optimizer.apply_gradients": [[462, null]], "mlx.optimizers.Optimizer.init": [[463, null]], "mlx.optimizers.Optimizer.state": [[464, null]], "mlx.optimizers.Optimizer.update": [[465, null]], "mlx.optimizers.RMSprop": [[466, null]], "mlx.optimizers.SGD": [[467, null]], "mlx.optimizers.clip_grad_norm": [[309, null]], "mlx.optimizers.cosine_decay": [[468, null]], "mlx.optimizers.exponential_decay": [[469, null]], "mlx.optimizers.join_schedules": [[470, null]], "mlx.optimizers.linear_schedule": [[471, null]], "mlx.optimizers.step_decay": [[472, null]], "mlx.utils.tree_flatten": [[310, null]], "mlx.utils.tree_map": [[311, null]], "mlx.utils.tree_map_with_path": [[312, null]], "mlx.utils.tree_reduce": [[313, null]], "mlx.utils.tree_unflatten": [[314, null]], "x86 Shell": [[8, "x86-shell"]]}, "docnames": ["cpp/ops", "dev/custom_metal_kernels", "dev/extensions", "dev/metal_debugger", "examples/linear_regression", "examples/llama-inference", "examples/mlp", "index", "install", "python/_autosummary/mlx.core.Device", "python/_autosummary/mlx.core.Dtype", "python/_autosummary/mlx.core.DtypeCategory", "python/_autosummary/mlx.core.abs", "python/_autosummary/mlx.core.add", "python/_autosummary/mlx.core.addmm", "python/_autosummary/mlx.core.all", "python/_autosummary/mlx.core.allclose", "python/_autosummary/mlx.core.any", "python/_autosummary/mlx.core.arange", "python/_autosummary/mlx.core.arccos", "python/_autosummary/mlx.core.arccosh", "python/_autosummary/mlx.core.arcsin", "python/_autosummary/mlx.core.arcsinh", "python/_autosummary/mlx.core.arctan", "python/_autosummary/mlx.core.arctan2", "python/_autosummary/mlx.core.arctanh", "python/_autosummary/mlx.core.argmax", "python/_autosummary/mlx.core.argmin", "python/_autosummary/mlx.core.argpartition", "python/_autosummary/mlx.core.argsort", "python/_autosummary/mlx.core.array", "python/_autosummary/mlx.core.array.T", "python/_autosummary/mlx.core.array.abs", "python/_autosummary/mlx.core.array.all", "python/_autosummary/mlx.core.array.any", "python/_autosummary/mlx.core.array.argmax", "python/_autosummary/mlx.core.array.argmin", "python/_autosummary/mlx.core.array.astype", "python/_autosummary/mlx.core.array.at", "python/_autosummary/mlx.core.array.conj", "python/_autosummary/mlx.core.array.cos", "python/_autosummary/mlx.core.array.cummax", "python/_autosummary/mlx.core.array.cummin", "python/_autosummary/mlx.core.array.cumprod", "python/_autosummary/mlx.core.array.cumsum", "python/_autosummary/mlx.core.array.diag", "python/_autosummary/mlx.core.array.diagonal", "python/_autosummary/mlx.core.array.dtype", "python/_autosummary/mlx.core.array.exp", "python/_autosummary/mlx.core.array.flatten", "python/_autosummary/mlx.core.array.item", "python/_autosummary/mlx.core.array.itemsize", "python/_autosummary/mlx.core.array.log", "python/_autosummary/mlx.core.array.log10", "python/_autosummary/mlx.core.array.log1p", "python/_autosummary/mlx.core.array.log2", "python/_autosummary/mlx.core.array.logsumexp", "python/_autosummary/mlx.core.array.max", "python/_autosummary/mlx.core.array.mean", "python/_autosummary/mlx.core.array.min", "python/_autosummary/mlx.core.array.moveaxis", "python/_autosummary/mlx.core.array.nbytes", "python/_autosummary/mlx.core.array.ndim", "python/_autosummary/mlx.core.array.prod", "python/_autosummary/mlx.core.array.reciprocal", "python/_autosummary/mlx.core.array.reshape", "python/_autosummary/mlx.core.array.round", "python/_autosummary/mlx.core.array.rsqrt", "python/_autosummary/mlx.core.array.shape", "python/_autosummary/mlx.core.array.sin", "python/_autosummary/mlx.core.array.size", "python/_autosummary/mlx.core.array.split", "python/_autosummary/mlx.core.array.sqrt", "python/_autosummary/mlx.core.array.square", "python/_autosummary/mlx.core.array.squeeze", "python/_autosummary/mlx.core.array.std", "python/_autosummary/mlx.core.array.sum", "python/_autosummary/mlx.core.array.swapaxes", "python/_autosummary/mlx.core.array.tolist", "python/_autosummary/mlx.core.array.transpose", "python/_autosummary/mlx.core.array.var", "python/_autosummary/mlx.core.array.view", "python/_autosummary/mlx.core.array_equal", "python/_autosummary/mlx.core.as_strided", "python/_autosummary/mlx.core.atleast_1d", "python/_autosummary/mlx.core.atleast_2d", "python/_autosummary/mlx.core.atleast_3d", "python/_autosummary/mlx.core.bitwise_and", "python/_autosummary/mlx.core.bitwise_or", "python/_autosummary/mlx.core.bitwise_xor", "python/_autosummary/mlx.core.block_masked_mm", "python/_autosummary/mlx.core.broadcast_to", "python/_autosummary/mlx.core.ceil", "python/_autosummary/mlx.core.clip", "python/_autosummary/mlx.core.compile", "python/_autosummary/mlx.core.concatenate", "python/_autosummary/mlx.core.conj", "python/_autosummary/mlx.core.conjugate", "python/_autosummary/mlx.core.conv1d", "python/_autosummary/mlx.core.conv2d", "python/_autosummary/mlx.core.conv3d", "python/_autosummary/mlx.core.conv_general", "python/_autosummary/mlx.core.conv_transpose1d", "python/_autosummary/mlx.core.conv_transpose2d", "python/_autosummary/mlx.core.conv_transpose3d", "python/_autosummary/mlx.core.convolve", "python/_autosummary/mlx.core.cos", "python/_autosummary/mlx.core.cosh", "python/_autosummary/mlx.core.cummax", "python/_autosummary/mlx.core.cummin", "python/_autosummary/mlx.core.cumprod", "python/_autosummary/mlx.core.cumsum", "python/_autosummary/mlx.core.custom_function", "python/_autosummary/mlx.core.default_device", "python/_autosummary/mlx.core.default_stream", "python/_autosummary/mlx.core.degrees", "python/_autosummary/mlx.core.dequantize", "python/_autosummary/mlx.core.diag", "python/_autosummary/mlx.core.diagonal", "python/_autosummary/mlx.core.disable_compile", "python/_autosummary/mlx.core.distributed.Group", "python/_autosummary/mlx.core.distributed.all_gather", "python/_autosummary/mlx.core.distributed.all_sum", "python/_autosummary/mlx.core.distributed.init", "python/_autosummary/mlx.core.distributed.is_available", "python/_autosummary/mlx.core.distributed.recv", "python/_autosummary/mlx.core.distributed.recv_like", "python/_autosummary/mlx.core.distributed.send", "python/_autosummary/mlx.core.divide", "python/_autosummary/mlx.core.divmod", "python/_autosummary/mlx.core.einsum", "python/_autosummary/mlx.core.einsum_path", "python/_autosummary/mlx.core.enable_compile", "python/_autosummary/mlx.core.equal", "python/_autosummary/mlx.core.erf", "python/_autosummary/mlx.core.erfinv", "python/_autosummary/mlx.core.eval", "python/_autosummary/mlx.core.exp", "python/_autosummary/mlx.core.expand_dims", "python/_autosummary/mlx.core.expm1", "python/_autosummary/mlx.core.eye", "python/_autosummary/mlx.core.fast.affine_quantize", "python/_autosummary/mlx.core.fast.layer_norm", "python/_autosummary/mlx.core.fast.metal_kernel", "python/_autosummary/mlx.core.fast.rms_norm", "python/_autosummary/mlx.core.fast.rope", "python/_autosummary/mlx.core.fast.scaled_dot_product_attention", "python/_autosummary/mlx.core.fft.fft", "python/_autosummary/mlx.core.fft.fft2", "python/_autosummary/mlx.core.fft.fftn", "python/_autosummary/mlx.core.fft.ifft", "python/_autosummary/mlx.core.fft.ifft2", "python/_autosummary/mlx.core.fft.ifftn", "python/_autosummary/mlx.core.fft.irfft", "python/_autosummary/mlx.core.fft.irfft2", "python/_autosummary/mlx.core.fft.irfftn", "python/_autosummary/mlx.core.fft.rfft", "python/_autosummary/mlx.core.fft.rfft2", "python/_autosummary/mlx.core.fft.rfftn", "python/_autosummary/mlx.core.flatten", "python/_autosummary/mlx.core.floor", "python/_autosummary/mlx.core.floor_divide", "python/_autosummary/mlx.core.full", "python/_autosummary/mlx.core.gather_mm", "python/_autosummary/mlx.core.gather_qmm", "python/_autosummary/mlx.core.grad", "python/_autosummary/mlx.core.greater", "python/_autosummary/mlx.core.greater_equal", "python/_autosummary/mlx.core.hadamard_transform", "python/_autosummary/mlx.core.identity", "python/_autosummary/mlx.core.imag", "python/_autosummary/mlx.core.inner", "python/_autosummary/mlx.core.isclose", "python/_autosummary/mlx.core.isfinite", "python/_autosummary/mlx.core.isinf", "python/_autosummary/mlx.core.isnan", "python/_autosummary/mlx.core.isneginf", "python/_autosummary/mlx.core.isposinf", "python/_autosummary/mlx.core.issubdtype", "python/_autosummary/mlx.core.jvp", "python/_autosummary/mlx.core.left_shift", "python/_autosummary/mlx.core.less", "python/_autosummary/mlx.core.less_equal", "python/_autosummary/mlx.core.linalg.cholesky", "python/_autosummary/mlx.core.linalg.cholesky_inv", "python/_autosummary/mlx.core.linalg.cross", "python/_autosummary/mlx.core.linalg.eigh", "python/_autosummary/mlx.core.linalg.eigvalsh", "python/_autosummary/mlx.core.linalg.inv", "python/_autosummary/mlx.core.linalg.norm", "python/_autosummary/mlx.core.linalg.qr", "python/_autosummary/mlx.core.linalg.svd", "python/_autosummary/mlx.core.linalg.tri_inv", "python/_autosummary/mlx.core.linspace", "python/_autosummary/mlx.core.load", "python/_autosummary/mlx.core.log", "python/_autosummary/mlx.core.log10", "python/_autosummary/mlx.core.log1p", "python/_autosummary/mlx.core.log2", "python/_autosummary/mlx.core.logaddexp", "python/_autosummary/mlx.core.logical_and", "python/_autosummary/mlx.core.logical_not", "python/_autosummary/mlx.core.logical_or", "python/_autosummary/mlx.core.logsumexp", "python/_autosummary/mlx.core.matmul", "python/_autosummary/mlx.core.max", "python/_autosummary/mlx.core.maximum", "python/_autosummary/mlx.core.mean", "python/_autosummary/mlx.core.meshgrid", "python/_autosummary/mlx.core.metal.clear_cache", "python/_autosummary/mlx.core.metal.device_info", "python/_autosummary/mlx.core.metal.get_active_memory", "python/_autosummary/mlx.core.metal.get_cache_memory", "python/_autosummary/mlx.core.metal.get_peak_memory", "python/_autosummary/mlx.core.metal.is_available", "python/_autosummary/mlx.core.metal.reset_peak_memory", "python/_autosummary/mlx.core.metal.set_cache_limit", "python/_autosummary/mlx.core.metal.set_memory_limit", "python/_autosummary/mlx.core.metal.set_wired_limit", "python/_autosummary/mlx.core.metal.start_capture", "python/_autosummary/mlx.core.metal.stop_capture", "python/_autosummary/mlx.core.min", "python/_autosummary/mlx.core.minimum", "python/_autosummary/mlx.core.moveaxis", "python/_autosummary/mlx.core.multiply", "python/_autosummary/mlx.core.nan_to_num", "python/_autosummary/mlx.core.negative", "python/_autosummary/mlx.core.new_stream", "python/_autosummary/mlx.core.not_equal", "python/_autosummary/mlx.core.ones", "python/_autosummary/mlx.core.ones_like", "python/_autosummary/mlx.core.outer", "python/_autosummary/mlx.core.pad", "python/_autosummary/mlx.core.partition", "python/_autosummary/mlx.core.power", "python/_autosummary/mlx.core.prod", "python/_autosummary/mlx.core.put_along_axis", "python/_autosummary/mlx.core.quantize", "python/_autosummary/mlx.core.quantized_matmul", "python/_autosummary/mlx.core.radians", "python/_autosummary/mlx.core.random.bernoulli", "python/_autosummary/mlx.core.random.categorical", "python/_autosummary/mlx.core.random.gumbel", "python/_autosummary/mlx.core.random.key", "python/_autosummary/mlx.core.random.laplace", "python/_autosummary/mlx.core.random.multivariate_normal", "python/_autosummary/mlx.core.random.normal", "python/_autosummary/mlx.core.random.permutation", "python/_autosummary/mlx.core.random.randint", "python/_autosummary/mlx.core.random.seed", "python/_autosummary/mlx.core.random.split", "python/_autosummary/mlx.core.random.truncated_normal", "python/_autosummary/mlx.core.random.uniform", "python/_autosummary/mlx.core.real", "python/_autosummary/mlx.core.reciprocal", "python/_autosummary/mlx.core.remainder", "python/_autosummary/mlx.core.repeat", "python/_autosummary/mlx.core.reshape", "python/_autosummary/mlx.core.right_shift", "python/_autosummary/mlx.core.roll", "python/_autosummary/mlx.core.round", "python/_autosummary/mlx.core.rsqrt", "python/_autosummary/mlx.core.save", "python/_autosummary/mlx.core.save_gguf", "python/_autosummary/mlx.core.save_safetensors", "python/_autosummary/mlx.core.savez", "python/_autosummary/mlx.core.savez_compressed", "python/_autosummary/mlx.core.set_default_device", "python/_autosummary/mlx.core.set_default_stream", "python/_autosummary/mlx.core.sigmoid", "python/_autosummary/mlx.core.sign", "python/_autosummary/mlx.core.sin", "python/_autosummary/mlx.core.sinh", "python/_autosummary/mlx.core.softmax", "python/_autosummary/mlx.core.sort", "python/_autosummary/mlx.core.split", "python/_autosummary/mlx.core.sqrt", "python/_autosummary/mlx.core.square", "python/_autosummary/mlx.core.squeeze", "python/_autosummary/mlx.core.stack", "python/_autosummary/mlx.core.std", "python/_autosummary/mlx.core.stop_gradient", "python/_autosummary/mlx.core.stream", "python/_autosummary/mlx.core.subtract", "python/_autosummary/mlx.core.sum", "python/_autosummary/mlx.core.swapaxes", "python/_autosummary/mlx.core.synchronize", "python/_autosummary/mlx.core.take", "python/_autosummary/mlx.core.take_along_axis", "python/_autosummary/mlx.core.tan", "python/_autosummary/mlx.core.tanh", "python/_autosummary/mlx.core.tensordot", "python/_autosummary/mlx.core.tile", "python/_autosummary/mlx.core.topk", "python/_autosummary/mlx.core.trace", "python/_autosummary/mlx.core.transpose", "python/_autosummary/mlx.core.tri", "python/_autosummary/mlx.core.tril", "python/_autosummary/mlx.core.triu", "python/_autosummary/mlx.core.value_and_grad", "python/_autosummary/mlx.core.var", "python/_autosummary/mlx.core.view", "python/_autosummary/mlx.core.vjp", "python/_autosummary/mlx.core.vmap", "python/_autosummary/mlx.core.where", "python/_autosummary/mlx.core.zeros", "python/_autosummary/mlx.core.zeros_like", "python/_autosummary/mlx.nn.quantize", "python/_autosummary/mlx.nn.value_and_grad", "python/_autosummary/mlx.optimizers.clip_grad_norm", "python/_autosummary/mlx.utils.tree_flatten", "python/_autosummary/mlx.utils.tree_map", "python/_autosummary/mlx.utils.tree_map_with_path", "python/_autosummary/mlx.utils.tree_reduce", "python/_autosummary/mlx.utils.tree_unflatten", "python/_autosummary/stream_class", "python/array", "python/data_types", "python/devices_and_streams", "python/distributed", "python/fast", "python/fft", "python/linalg", "python/metal", "python/nn", "python/nn/_autosummary/mlx.nn.ALiBi", "python/nn/_autosummary/mlx.nn.AvgPool1d", "python/nn/_autosummary/mlx.nn.AvgPool2d", "python/nn/_autosummary/mlx.nn.BatchNorm", "python/nn/_autosummary/mlx.nn.CELU", "python/nn/_autosummary/mlx.nn.Conv1d", "python/nn/_autosummary/mlx.nn.Conv2d", "python/nn/_autosummary/mlx.nn.Conv3d", "python/nn/_autosummary/mlx.nn.ConvTranspose1d", "python/nn/_autosummary/mlx.nn.ConvTranspose2d", "python/nn/_autosummary/mlx.nn.ConvTranspose3d", "python/nn/_autosummary/mlx.nn.Dropout", "python/nn/_autosummary/mlx.nn.Dropout2d", "python/nn/_autosummary/mlx.nn.Dropout3d", "python/nn/_autosummary/mlx.nn.ELU", "python/nn/_autosummary/mlx.nn.Embedding", "python/nn/_autosummary/mlx.nn.GELU", "python/nn/_autosummary/mlx.nn.GLU", "python/nn/_autosummary/mlx.nn.GRU", "python/nn/_autosummary/mlx.nn.GroupNorm", "python/nn/_autosummary/mlx.nn.HardShrink", "python/nn/_autosummary/mlx.nn.HardTanh", "python/nn/_autosummary/mlx.nn.Hardswish", "python/nn/_autosummary/mlx.nn.InstanceNorm", "python/nn/_autosummary/mlx.nn.LSTM", "python/nn/_autosummary/mlx.nn.LayerNorm", "python/nn/_autosummary/mlx.nn.LeakyReLU", "python/nn/_autosummary/mlx.nn.Linear", "python/nn/_autosummary/mlx.nn.LogSigmoid", "python/nn/_autosummary/mlx.nn.LogSoftmax", "python/nn/_autosummary/mlx.nn.MaxPool1d", "python/nn/_autosummary/mlx.nn.MaxPool2d", "python/nn/_autosummary/mlx.nn.Mish", "python/nn/_autosummary/mlx.nn.Module.apply", "python/nn/_autosummary/mlx.nn.Module.apply_to_modules", "python/nn/_autosummary/mlx.nn.Module.children", "python/nn/_autosummary/mlx.nn.Module.eval", "python/nn/_autosummary/mlx.nn.Module.filter_and_map", "python/nn/_autosummary/mlx.nn.Module.freeze", "python/nn/_autosummary/mlx.nn.Module.leaf_modules", "python/nn/_autosummary/mlx.nn.Module.load_weights", "python/nn/_autosummary/mlx.nn.Module.modules", "python/nn/_autosummary/mlx.nn.Module.named_modules", "python/nn/_autosummary/mlx.nn.Module.parameters", "python/nn/_autosummary/mlx.nn.Module.save_weights", "python/nn/_autosummary/mlx.nn.Module.set_dtype", "python/nn/_autosummary/mlx.nn.Module.state", "python/nn/_autosummary/mlx.nn.Module.train", "python/nn/_autosummary/mlx.nn.Module.trainable_parameters", "python/nn/_autosummary/mlx.nn.Module.training", "python/nn/_autosummary/mlx.nn.Module.unfreeze", "python/nn/_autosummary/mlx.nn.Module.update", "python/nn/_autosummary/mlx.nn.Module.update_modules", "python/nn/_autosummary/mlx.nn.MultiHeadAttention", "python/nn/_autosummary/mlx.nn.PReLU", "python/nn/_autosummary/mlx.nn.QuantizedEmbedding", "python/nn/_autosummary/mlx.nn.QuantizedLinear", "python/nn/_autosummary/mlx.nn.RMSNorm", "python/nn/_autosummary/mlx.nn.RNN", "python/nn/_autosummary/mlx.nn.ReLU", "python/nn/_autosummary/mlx.nn.ReLU6", "python/nn/_autosummary/mlx.nn.RoPE", "python/nn/_autosummary/mlx.nn.SELU", "python/nn/_autosummary/mlx.nn.Sequential", "python/nn/_autosummary/mlx.nn.SiLU", "python/nn/_autosummary/mlx.nn.Sigmoid", "python/nn/_autosummary/mlx.nn.SinusoidalPositionalEncoding", "python/nn/_autosummary/mlx.nn.Softmax", "python/nn/_autosummary/mlx.nn.Softmin", "python/nn/_autosummary/mlx.nn.Softplus", "python/nn/_autosummary/mlx.nn.Softshrink", "python/nn/_autosummary/mlx.nn.Softsign", "python/nn/_autosummary/mlx.nn.Step", "python/nn/_autosummary/mlx.nn.Tanh", "python/nn/_autosummary/mlx.nn.Transformer", "python/nn/_autosummary/mlx.nn.Upsample", "python/nn/_autosummary/mlx.nn.init.constant", "python/nn/_autosummary/mlx.nn.init.glorot_normal", "python/nn/_autosummary/mlx.nn.init.glorot_uniform", "python/nn/_autosummary/mlx.nn.init.he_normal", "python/nn/_autosummary/mlx.nn.init.he_uniform", "python/nn/_autosummary/mlx.nn.init.identity", "python/nn/_autosummary/mlx.nn.init.normal", "python/nn/_autosummary/mlx.nn.init.uniform", "python/nn/_autosummary_functions/mlx.nn.celu", "python/nn/_autosummary_functions/mlx.nn.elu", "python/nn/_autosummary_functions/mlx.nn.gelu", "python/nn/_autosummary_functions/mlx.nn.gelu_approx", "python/nn/_autosummary_functions/mlx.nn.gelu_fast_approx", "python/nn/_autosummary_functions/mlx.nn.glu", "python/nn/_autosummary_functions/mlx.nn.hard_shrink", "python/nn/_autosummary_functions/mlx.nn.hard_tanh", "python/nn/_autosummary_functions/mlx.nn.hardswish", "python/nn/_autosummary_functions/mlx.nn.leaky_relu", "python/nn/_autosummary_functions/mlx.nn.log_sigmoid", "python/nn/_autosummary_functions/mlx.nn.log_softmax", "python/nn/_autosummary_functions/mlx.nn.losses.binary_cross_entropy", "python/nn/_autosummary_functions/mlx.nn.losses.cosine_similarity_loss", "python/nn/_autosummary_functions/mlx.nn.losses.cross_entropy", "python/nn/_autosummary_functions/mlx.nn.losses.gaussian_nll_loss", "python/nn/_autosummary_functions/mlx.nn.losses.hinge_loss", "python/nn/_autosummary_functions/mlx.nn.losses.huber_loss", "python/nn/_autosummary_functions/mlx.nn.losses.kl_div_loss", "python/nn/_autosummary_functions/mlx.nn.losses.l1_loss", "python/nn/_autosummary_functions/mlx.nn.losses.log_cosh_loss", "python/nn/_autosummary_functions/mlx.nn.losses.margin_ranking_loss", "python/nn/_autosummary_functions/mlx.nn.losses.mse_loss", "python/nn/_autosummary_functions/mlx.nn.losses.nll_loss", "python/nn/_autosummary_functions/mlx.nn.losses.smooth_l1_loss", "python/nn/_autosummary_functions/mlx.nn.losses.triplet_loss", "python/nn/_autosummary_functions/mlx.nn.mish", "python/nn/_autosummary_functions/mlx.nn.prelu", "python/nn/_autosummary_functions/mlx.nn.relu", "python/nn/_autosummary_functions/mlx.nn.relu6", "python/nn/_autosummary_functions/mlx.nn.selu", "python/nn/_autosummary_functions/mlx.nn.sigmoid", "python/nn/_autosummary_functions/mlx.nn.silu", "python/nn/_autosummary_functions/mlx.nn.softmax", "python/nn/_autosummary_functions/mlx.nn.softmin", "python/nn/_autosummary_functions/mlx.nn.softplus", "python/nn/_autosummary_functions/mlx.nn.softshrink", "python/nn/_autosummary_functions/mlx.nn.step", "python/nn/_autosummary_functions/mlx.nn.tanh", "python/nn/functions", "python/nn/init", "python/nn/layers", "python/nn/losses", "python/nn/module", "python/ops", "python/optimizers", "python/optimizers/_autosummary/mlx.optimizers.AdaDelta", "python/optimizers/_autosummary/mlx.optimizers.Adafactor", "python/optimizers/_autosummary/mlx.optimizers.Adagrad", "python/optimizers/_autosummary/mlx.optimizers.Adam", "python/optimizers/_autosummary/mlx.optimizers.AdamW", "python/optimizers/_autosummary/mlx.optimizers.Adamax", "python/optimizers/_autosummary/mlx.optimizers.Lion", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.apply_gradients", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.init", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.state", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.update", "python/optimizers/_autosummary/mlx.optimizers.RMSprop", "python/optimizers/_autosummary/mlx.optimizers.SGD", "python/optimizers/_autosummary/mlx.optimizers.cosine_decay", "python/optimizers/_autosummary/mlx.optimizers.exponential_decay", "python/optimizers/_autosummary/mlx.optimizers.join_schedules", "python/optimizers/_autosummary/mlx.optimizers.linear_schedule", "python/optimizers/_autosummary/mlx.optimizers.step_decay", "python/optimizers/common_optimizers", "python/optimizers/optimizer", "python/optimizers/schedulers", "python/random", "python/transforms", "python/tree_utils", "usage/compile", "usage/distributed", "usage/function_transforms", "usage/indexing", "usage/lazy_evaluation", "usage/numpy", "usage/quick_start", "usage/saving_and_loading", "usage/unified_memory", "usage/using_streams"], "envversion": {"sphinx": 62, "sphinx.domains.c": 3, "sphinx.domains.changeset": 1, "sphinx.domains.citation": 1, "sphinx.domains.cpp": 9, "sphinx.domains.index": 1, "sphinx.domains.javascript": 3, "sphinx.domains.math": 2, "sphinx.domains.python": 4, "sphinx.domains.rst": 2, "sphinx.domains.std": 2, "sphinx.ext.intersphinx": 1}, "filenames": ["cpp/ops.rst", "dev/custom_metal_kernels.rst", "dev/extensions.rst", "dev/metal_debugger.rst", "examples/linear_regression.rst", "examples/llama-inference.rst", "examples/mlp.rst", "index.rst", "install.rst", "python/_autosummary/mlx.core.Device.rst", "python/_autosummary/mlx.core.Dtype.rst", "python/_autosummary/mlx.core.DtypeCategory.rst", "python/_autosummary/mlx.core.abs.rst", "python/_autosummary/mlx.core.add.rst", "python/_autosummary/mlx.core.addmm.rst", "python/_autosummary/mlx.core.all.rst", "python/_autosummary/mlx.core.allclose.rst", "python/_autosummary/mlx.core.any.rst", "python/_autosummary/mlx.core.arange.rst", "python/_autosummary/mlx.core.arccos.rst", "python/_autosummary/mlx.core.arccosh.rst", "python/_autosummary/mlx.core.arcsin.rst", "python/_autosummary/mlx.core.arcsinh.rst", "python/_autosummary/mlx.core.arctan.rst", "python/_autosummary/mlx.core.arctan2.rst", "python/_autosummary/mlx.core.arctanh.rst", "python/_autosummary/mlx.core.argmax.rst", "python/_autosummary/mlx.core.argmin.rst", "python/_autosummary/mlx.core.argpartition.rst", "python/_autosummary/mlx.core.argsort.rst", "python/_autosummary/mlx.core.array.rst", "python/_autosummary/mlx.core.array.T.rst", "python/_autosummary/mlx.core.array.abs.rst", "python/_autosummary/mlx.core.array.all.rst", "python/_autosummary/mlx.core.array.any.rst", "python/_autosummary/mlx.core.array.argmax.rst", "python/_autosummary/mlx.core.array.argmin.rst", "python/_autosummary/mlx.core.array.astype.rst", "python/_autosummary/mlx.core.array.at.rst", "python/_autosummary/mlx.core.array.conj.rst", "python/_autosummary/mlx.core.array.cos.rst", "python/_autosummary/mlx.core.array.cummax.rst", "python/_autosummary/mlx.core.array.cummin.rst", "python/_autosummary/mlx.core.array.cumprod.rst", "python/_autosummary/mlx.core.array.cumsum.rst", "python/_autosummary/mlx.core.array.diag.rst", "python/_autosummary/mlx.core.array.diagonal.rst", "python/_autosummary/mlx.core.array.dtype.rst", "python/_autosummary/mlx.core.array.exp.rst", "python/_autosummary/mlx.core.array.flatten.rst", "python/_autosummary/mlx.core.array.item.rst", "python/_autosummary/mlx.core.array.itemsize.rst", "python/_autosummary/mlx.core.array.log.rst", "python/_autosummary/mlx.core.array.log10.rst", "python/_autosummary/mlx.core.array.log1p.rst", "python/_autosummary/mlx.core.array.log2.rst", "python/_autosummary/mlx.core.array.logsumexp.rst", "python/_autosummary/mlx.core.array.max.rst", "python/_autosummary/mlx.core.array.mean.rst", "python/_autosummary/mlx.core.array.min.rst", "python/_autosummary/mlx.core.array.moveaxis.rst", "python/_autosummary/mlx.core.array.nbytes.rst", "python/_autosummary/mlx.core.array.ndim.rst", "python/_autosummary/mlx.core.array.prod.rst", "python/_autosummary/mlx.core.array.reciprocal.rst", "python/_autosummary/mlx.core.array.reshape.rst", "python/_autosummary/mlx.core.array.round.rst", "python/_autosummary/mlx.core.array.rsqrt.rst", "python/_autosummary/mlx.core.array.shape.rst", "python/_autosummary/mlx.core.array.sin.rst", "python/_autosummary/mlx.core.array.size.rst", "python/_autosummary/mlx.core.array.split.rst", "python/_autosummary/mlx.core.array.sqrt.rst", "python/_autosummary/mlx.core.array.square.rst", "python/_autosummary/mlx.core.array.squeeze.rst", "python/_autosummary/mlx.core.array.std.rst", "python/_autosummary/mlx.core.array.sum.rst", "python/_autosummary/mlx.core.array.swapaxes.rst", "python/_autosummary/mlx.core.array.tolist.rst", "python/_autosummary/mlx.core.array.transpose.rst", "python/_autosummary/mlx.core.array.var.rst", "python/_autosummary/mlx.core.array.view.rst", "python/_autosummary/mlx.core.array_equal.rst", "python/_autosummary/mlx.core.as_strided.rst", "python/_autosummary/mlx.core.atleast_1d.rst", "python/_autosummary/mlx.core.atleast_2d.rst", "python/_autosummary/mlx.core.atleast_3d.rst", "python/_autosummary/mlx.core.bitwise_and.rst", "python/_autosummary/mlx.core.bitwise_or.rst", "python/_autosummary/mlx.core.bitwise_xor.rst", "python/_autosummary/mlx.core.block_masked_mm.rst", "python/_autosummary/mlx.core.broadcast_to.rst", "python/_autosummary/mlx.core.ceil.rst", "python/_autosummary/mlx.core.clip.rst", "python/_autosummary/mlx.core.compile.rst", "python/_autosummary/mlx.core.concatenate.rst", "python/_autosummary/mlx.core.conj.rst", "python/_autosummary/mlx.core.conjugate.rst", "python/_autosummary/mlx.core.conv1d.rst", "python/_autosummary/mlx.core.conv2d.rst", "python/_autosummary/mlx.core.conv3d.rst", "python/_autosummary/mlx.core.conv_general.rst", "python/_autosummary/mlx.core.conv_transpose1d.rst", "python/_autosummary/mlx.core.conv_transpose2d.rst", "python/_autosummary/mlx.core.conv_transpose3d.rst", "python/_autosummary/mlx.core.convolve.rst", "python/_autosummary/mlx.core.cos.rst", "python/_autosummary/mlx.core.cosh.rst", "python/_autosummary/mlx.core.cummax.rst", "python/_autosummary/mlx.core.cummin.rst", "python/_autosummary/mlx.core.cumprod.rst", "python/_autosummary/mlx.core.cumsum.rst", "python/_autosummary/mlx.core.custom_function.rst", "python/_autosummary/mlx.core.default_device.rst", "python/_autosummary/mlx.core.default_stream.rst", "python/_autosummary/mlx.core.degrees.rst", "python/_autosummary/mlx.core.dequantize.rst", "python/_autosummary/mlx.core.diag.rst", "python/_autosummary/mlx.core.diagonal.rst", "python/_autosummary/mlx.core.disable_compile.rst", "python/_autosummary/mlx.core.distributed.Group.rst", "python/_autosummary/mlx.core.distributed.all_gather.rst", "python/_autosummary/mlx.core.distributed.all_sum.rst", "python/_autosummary/mlx.core.distributed.init.rst", "python/_autosummary/mlx.core.distributed.is_available.rst", "python/_autosummary/mlx.core.distributed.recv.rst", "python/_autosummary/mlx.core.distributed.recv_like.rst", "python/_autosummary/mlx.core.distributed.send.rst", "python/_autosummary/mlx.core.divide.rst", "python/_autosummary/mlx.core.divmod.rst", "python/_autosummary/mlx.core.einsum.rst", "python/_autosummary/mlx.core.einsum_path.rst", "python/_autosummary/mlx.core.enable_compile.rst", "python/_autosummary/mlx.core.equal.rst", "python/_autosummary/mlx.core.erf.rst", "python/_autosummary/mlx.core.erfinv.rst", "python/_autosummary/mlx.core.eval.rst", "python/_autosummary/mlx.core.exp.rst", "python/_autosummary/mlx.core.expand_dims.rst", "python/_autosummary/mlx.core.expm1.rst", "python/_autosummary/mlx.core.eye.rst", "python/_autosummary/mlx.core.fast.affine_quantize.rst", "python/_autosummary/mlx.core.fast.layer_norm.rst", "python/_autosummary/mlx.core.fast.metal_kernel.rst", "python/_autosummary/mlx.core.fast.rms_norm.rst", "python/_autosummary/mlx.core.fast.rope.rst", "python/_autosummary/mlx.core.fast.scaled_dot_product_attention.rst", "python/_autosummary/mlx.core.fft.fft.rst", "python/_autosummary/mlx.core.fft.fft2.rst", "python/_autosummary/mlx.core.fft.fftn.rst", "python/_autosummary/mlx.core.fft.ifft.rst", "python/_autosummary/mlx.core.fft.ifft2.rst", "python/_autosummary/mlx.core.fft.ifftn.rst", "python/_autosummary/mlx.core.fft.irfft.rst", "python/_autosummary/mlx.core.fft.irfft2.rst", "python/_autosummary/mlx.core.fft.irfftn.rst", "python/_autosummary/mlx.core.fft.rfft.rst", "python/_autosummary/mlx.core.fft.rfft2.rst", "python/_autosummary/mlx.core.fft.rfftn.rst", "python/_autosummary/mlx.core.flatten.rst", "python/_autosummary/mlx.core.floor.rst", "python/_autosummary/mlx.core.floor_divide.rst", "python/_autosummary/mlx.core.full.rst", "python/_autosummary/mlx.core.gather_mm.rst", "python/_autosummary/mlx.core.gather_qmm.rst", "python/_autosummary/mlx.core.grad.rst", "python/_autosummary/mlx.core.greater.rst", "python/_autosummary/mlx.core.greater_equal.rst", "python/_autosummary/mlx.core.hadamard_transform.rst", "python/_autosummary/mlx.core.identity.rst", "python/_autosummary/mlx.core.imag.rst", "python/_autosummary/mlx.core.inner.rst", "python/_autosummary/mlx.core.isclose.rst", "python/_autosummary/mlx.core.isfinite.rst", "python/_autosummary/mlx.core.isinf.rst", "python/_autosummary/mlx.core.isnan.rst", "python/_autosummary/mlx.core.isneginf.rst", "python/_autosummary/mlx.core.isposinf.rst", "python/_autosummary/mlx.core.issubdtype.rst", "python/_autosummary/mlx.core.jvp.rst", "python/_autosummary/mlx.core.left_shift.rst", "python/_autosummary/mlx.core.less.rst", "python/_autosummary/mlx.core.less_equal.rst", "python/_autosummary/mlx.core.linalg.cholesky.rst", "python/_autosummary/mlx.core.linalg.cholesky_inv.rst", "python/_autosummary/mlx.core.linalg.cross.rst", "python/_autosummary/mlx.core.linalg.eigh.rst", "python/_autosummary/mlx.core.linalg.eigvalsh.rst", "python/_autosummary/mlx.core.linalg.inv.rst", "python/_autosummary/mlx.core.linalg.norm.rst", "python/_autosummary/mlx.core.linalg.qr.rst", "python/_autosummary/mlx.core.linalg.svd.rst", "python/_autosummary/mlx.core.linalg.tri_inv.rst", "python/_autosummary/mlx.core.linspace.rst", "python/_autosummary/mlx.core.load.rst", "python/_autosummary/mlx.core.log.rst", "python/_autosummary/mlx.core.log10.rst", "python/_autosummary/mlx.core.log1p.rst", "python/_autosummary/mlx.core.log2.rst", "python/_autosummary/mlx.core.logaddexp.rst", "python/_autosummary/mlx.core.logical_and.rst", "python/_autosummary/mlx.core.logical_not.rst", "python/_autosummary/mlx.core.logical_or.rst", "python/_autosummary/mlx.core.logsumexp.rst", "python/_autosummary/mlx.core.matmul.rst", "python/_autosummary/mlx.core.max.rst", "python/_autosummary/mlx.core.maximum.rst", "python/_autosummary/mlx.core.mean.rst", "python/_autosummary/mlx.core.meshgrid.rst", "python/_autosummary/mlx.core.metal.clear_cache.rst", "python/_autosummary/mlx.core.metal.device_info.rst", "python/_autosummary/mlx.core.metal.get_active_memory.rst", "python/_autosummary/mlx.core.metal.get_cache_memory.rst", "python/_autosummary/mlx.core.metal.get_peak_memory.rst", "python/_autosummary/mlx.core.metal.is_available.rst", "python/_autosummary/mlx.core.metal.reset_peak_memory.rst", "python/_autosummary/mlx.core.metal.set_cache_limit.rst", "python/_autosummary/mlx.core.metal.set_memory_limit.rst", "python/_autosummary/mlx.core.metal.set_wired_limit.rst", "python/_autosummary/mlx.core.metal.start_capture.rst", "python/_autosummary/mlx.core.metal.stop_capture.rst", "python/_autosummary/mlx.core.min.rst", "python/_autosummary/mlx.core.minimum.rst", "python/_autosummary/mlx.core.moveaxis.rst", "python/_autosummary/mlx.core.multiply.rst", "python/_autosummary/mlx.core.nan_to_num.rst", "python/_autosummary/mlx.core.negative.rst", "python/_autosummary/mlx.core.new_stream.rst", "python/_autosummary/mlx.core.not_equal.rst", "python/_autosummary/mlx.core.ones.rst", "python/_autosummary/mlx.core.ones_like.rst", "python/_autosummary/mlx.core.outer.rst", "python/_autosummary/mlx.core.pad.rst", "python/_autosummary/mlx.core.partition.rst", "python/_autosummary/mlx.core.power.rst", "python/_autosummary/mlx.core.prod.rst", "python/_autosummary/mlx.core.put_along_axis.rst", "python/_autosummary/mlx.core.quantize.rst", "python/_autosummary/mlx.core.quantized_matmul.rst", "python/_autosummary/mlx.core.radians.rst", "python/_autosummary/mlx.core.random.bernoulli.rst", "python/_autosummary/mlx.core.random.categorical.rst", "python/_autosummary/mlx.core.random.gumbel.rst", "python/_autosummary/mlx.core.random.key.rst", "python/_autosummary/mlx.core.random.laplace.rst", "python/_autosummary/mlx.core.random.multivariate_normal.rst", "python/_autosummary/mlx.core.random.normal.rst", "python/_autosummary/mlx.core.random.permutation.rst", "python/_autosummary/mlx.core.random.randint.rst", "python/_autosummary/mlx.core.random.seed.rst", "python/_autosummary/mlx.core.random.split.rst", "python/_autosummary/mlx.core.random.truncated_normal.rst", "python/_autosummary/mlx.core.random.uniform.rst", "python/_autosummary/mlx.core.real.rst", "python/_autosummary/mlx.core.reciprocal.rst", "python/_autosummary/mlx.core.remainder.rst", "python/_autosummary/mlx.core.repeat.rst", "python/_autosummary/mlx.core.reshape.rst", "python/_autosummary/mlx.core.right_shift.rst", "python/_autosummary/mlx.core.roll.rst", "python/_autosummary/mlx.core.round.rst", "python/_autosummary/mlx.core.rsqrt.rst", "python/_autosummary/mlx.core.save.rst", "python/_autosummary/mlx.core.save_gguf.rst", "python/_autosummary/mlx.core.save_safetensors.rst", "python/_autosummary/mlx.core.savez.rst", "python/_autosummary/mlx.core.savez_compressed.rst", "python/_autosummary/mlx.core.set_default_device.rst", "python/_autosummary/mlx.core.set_default_stream.rst", "python/_autosummary/mlx.core.sigmoid.rst", "python/_autosummary/mlx.core.sign.rst", "python/_autosummary/mlx.core.sin.rst", "python/_autosummary/mlx.core.sinh.rst", "python/_autosummary/mlx.core.softmax.rst", "python/_autosummary/mlx.core.sort.rst", "python/_autosummary/mlx.core.split.rst", "python/_autosummary/mlx.core.sqrt.rst", "python/_autosummary/mlx.core.square.rst", "python/_autosummary/mlx.core.squeeze.rst", "python/_autosummary/mlx.core.stack.rst", "python/_autosummary/mlx.core.std.rst", "python/_autosummary/mlx.core.stop_gradient.rst", "python/_autosummary/mlx.core.stream.rst", "python/_autosummary/mlx.core.subtract.rst", "python/_autosummary/mlx.core.sum.rst", "python/_autosummary/mlx.core.swapaxes.rst", "python/_autosummary/mlx.core.synchronize.rst", "python/_autosummary/mlx.core.take.rst", "python/_autosummary/mlx.core.take_along_axis.rst", "python/_autosummary/mlx.core.tan.rst", "python/_autosummary/mlx.core.tanh.rst", "python/_autosummary/mlx.core.tensordot.rst", "python/_autosummary/mlx.core.tile.rst", "python/_autosummary/mlx.core.topk.rst", "python/_autosummary/mlx.core.trace.rst", "python/_autosummary/mlx.core.transpose.rst", "python/_autosummary/mlx.core.tri.rst", "python/_autosummary/mlx.core.tril.rst", "python/_autosummary/mlx.core.triu.rst", "python/_autosummary/mlx.core.value_and_grad.rst", "python/_autosummary/mlx.core.var.rst", "python/_autosummary/mlx.core.view.rst", "python/_autosummary/mlx.core.vjp.rst", "python/_autosummary/mlx.core.vmap.rst", "python/_autosummary/mlx.core.where.rst", "python/_autosummary/mlx.core.zeros.rst", "python/_autosummary/mlx.core.zeros_like.rst", "python/_autosummary/mlx.nn.quantize.rst", "python/_autosummary/mlx.nn.value_and_grad.rst", "python/_autosummary/mlx.optimizers.clip_grad_norm.rst", "python/_autosummary/mlx.utils.tree_flatten.rst", "python/_autosummary/mlx.utils.tree_map.rst", "python/_autosummary/mlx.utils.tree_map_with_path.rst", "python/_autosummary/mlx.utils.tree_reduce.rst", "python/_autosummary/mlx.utils.tree_unflatten.rst", "python/_autosummary/stream_class.rst", "python/array.rst", "python/data_types.rst", "python/devices_and_streams.rst", "python/distributed.rst", "python/fast.rst", "python/fft.rst", "python/linalg.rst", "python/metal.rst", "python/nn.rst", "python/nn/_autosummary/mlx.nn.ALiBi.rst", "python/nn/_autosummary/mlx.nn.AvgPool1d.rst", "python/nn/_autosummary/mlx.nn.AvgPool2d.rst", "python/nn/_autosummary/mlx.nn.BatchNorm.rst", "python/nn/_autosummary/mlx.nn.CELU.rst", "python/nn/_autosummary/mlx.nn.Conv1d.rst", "python/nn/_autosummary/mlx.nn.Conv2d.rst", "python/nn/_autosummary/mlx.nn.Conv3d.rst", "python/nn/_autosummary/mlx.nn.ConvTranspose1d.rst", "python/nn/_autosummary/mlx.nn.ConvTranspose2d.rst", "python/nn/_autosummary/mlx.nn.ConvTranspose3d.rst", "python/nn/_autosummary/mlx.nn.Dropout.rst", "python/nn/_autosummary/mlx.nn.Dropout2d.rst", "python/nn/_autosummary/mlx.nn.Dropout3d.rst", "python/nn/_autosummary/mlx.nn.ELU.rst", "python/nn/_autosummary/mlx.nn.Embedding.rst", "python/nn/_autosummary/mlx.nn.GELU.rst", "python/nn/_autosummary/mlx.nn.GLU.rst", "python/nn/_autosummary/mlx.nn.GRU.rst", "python/nn/_autosummary/mlx.nn.GroupNorm.rst", "python/nn/_autosummary/mlx.nn.HardShrink.rst", "python/nn/_autosummary/mlx.nn.HardTanh.rst", "python/nn/_autosummary/mlx.nn.Hardswish.rst", "python/nn/_autosummary/mlx.nn.InstanceNorm.rst", "python/nn/_autosummary/mlx.nn.LSTM.rst", "python/nn/_autosummary/mlx.nn.LayerNorm.rst", "python/nn/_autosummary/mlx.nn.LeakyReLU.rst", "python/nn/_autosummary/mlx.nn.Linear.rst", "python/nn/_autosummary/mlx.nn.LogSigmoid.rst", "python/nn/_autosummary/mlx.nn.LogSoftmax.rst", "python/nn/_autosummary/mlx.nn.MaxPool1d.rst", "python/nn/_autosummary/mlx.nn.MaxPool2d.rst", "python/nn/_autosummary/mlx.nn.Mish.rst", "python/nn/_autosummary/mlx.nn.Module.apply.rst", "python/nn/_autosummary/mlx.nn.Module.apply_to_modules.rst", "python/nn/_autosummary/mlx.nn.Module.children.rst", "python/nn/_autosummary/mlx.nn.Module.eval.rst", "python/nn/_autosummary/mlx.nn.Module.filter_and_map.rst", "python/nn/_autosummary/mlx.nn.Module.freeze.rst", "python/nn/_autosummary/mlx.nn.Module.leaf_modules.rst", "python/nn/_autosummary/mlx.nn.Module.load_weights.rst", "python/nn/_autosummary/mlx.nn.Module.modules.rst", "python/nn/_autosummary/mlx.nn.Module.named_modules.rst", "python/nn/_autosummary/mlx.nn.Module.parameters.rst", "python/nn/_autosummary/mlx.nn.Module.save_weights.rst", "python/nn/_autosummary/mlx.nn.Module.set_dtype.rst", "python/nn/_autosummary/mlx.nn.Module.state.rst", "python/nn/_autosummary/mlx.nn.Module.train.rst", "python/nn/_autosummary/mlx.nn.Module.trainable_parameters.rst", "python/nn/_autosummary/mlx.nn.Module.training.rst", "python/nn/_autosummary/mlx.nn.Module.unfreeze.rst", "python/nn/_autosummary/mlx.nn.Module.update.rst", "python/nn/_autosummary/mlx.nn.Module.update_modules.rst", "python/nn/_autosummary/mlx.nn.MultiHeadAttention.rst", "python/nn/_autosummary/mlx.nn.PReLU.rst", "python/nn/_autosummary/mlx.nn.QuantizedEmbedding.rst", "python/nn/_autosummary/mlx.nn.QuantizedLinear.rst", "python/nn/_autosummary/mlx.nn.RMSNorm.rst", "python/nn/_autosummary/mlx.nn.RNN.rst", "python/nn/_autosummary/mlx.nn.ReLU.rst", "python/nn/_autosummary/mlx.nn.ReLU6.rst", "python/nn/_autosummary/mlx.nn.RoPE.rst", "python/nn/_autosummary/mlx.nn.SELU.rst", "python/nn/_autosummary/mlx.nn.Sequential.rst", "python/nn/_autosummary/mlx.nn.SiLU.rst", "python/nn/_autosummary/mlx.nn.Sigmoid.rst", "python/nn/_autosummary/mlx.nn.SinusoidalPositionalEncoding.rst", "python/nn/_autosummary/mlx.nn.Softmax.rst", "python/nn/_autosummary/mlx.nn.Softmin.rst", "python/nn/_autosummary/mlx.nn.Softplus.rst", "python/nn/_autosummary/mlx.nn.Softshrink.rst", "python/nn/_autosummary/mlx.nn.Softsign.rst", "python/nn/_autosummary/mlx.nn.Step.rst", "python/nn/_autosummary/mlx.nn.Tanh.rst", "python/nn/_autosummary/mlx.nn.Transformer.rst", "python/nn/_autosummary/mlx.nn.Upsample.rst", "python/nn/_autosummary/mlx.nn.init.constant.rst", "python/nn/_autosummary/mlx.nn.init.glorot_normal.rst", "python/nn/_autosummary/mlx.nn.init.glorot_uniform.rst", "python/nn/_autosummary/mlx.nn.init.he_normal.rst", "python/nn/_autosummary/mlx.nn.init.he_uniform.rst", "python/nn/_autosummary/mlx.nn.init.identity.rst", "python/nn/_autosummary/mlx.nn.init.normal.rst", "python/nn/_autosummary/mlx.nn.init.uniform.rst", "python/nn/_autosummary_functions/mlx.nn.celu.rst", "python/nn/_autosummary_functions/mlx.nn.elu.rst", "python/nn/_autosummary_functions/mlx.nn.gelu.rst", "python/nn/_autosummary_functions/mlx.nn.gelu_approx.rst", "python/nn/_autosummary_functions/mlx.nn.gelu_fast_approx.rst", "python/nn/_autosummary_functions/mlx.nn.glu.rst", "python/nn/_autosummary_functions/mlx.nn.hard_shrink.rst", "python/nn/_autosummary_functions/mlx.nn.hard_tanh.rst", "python/nn/_autosummary_functions/mlx.nn.hardswish.rst", "python/nn/_autosummary_functions/mlx.nn.leaky_relu.rst", "python/nn/_autosummary_functions/mlx.nn.log_sigmoid.rst", "python/nn/_autosummary_functions/mlx.nn.log_softmax.rst", "python/nn/_autosummary_functions/mlx.nn.losses.binary_cross_entropy.rst", "python/nn/_autosummary_functions/mlx.nn.losses.cosine_similarity_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.cross_entropy.rst", "python/nn/_autosummary_functions/mlx.nn.losses.gaussian_nll_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.hinge_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.huber_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.kl_div_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.l1_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.log_cosh_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.margin_ranking_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.mse_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.nll_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.smooth_l1_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.triplet_loss.rst", "python/nn/_autosummary_functions/mlx.nn.mish.rst", "python/nn/_autosummary_functions/mlx.nn.prelu.rst", "python/nn/_autosummary_functions/mlx.nn.relu.rst", "python/nn/_autosummary_functions/mlx.nn.relu6.rst", "python/nn/_autosummary_functions/mlx.nn.selu.rst", "python/nn/_autosummary_functions/mlx.nn.sigmoid.rst", "python/nn/_autosummary_functions/mlx.nn.silu.rst", "python/nn/_autosummary_functions/mlx.nn.softmax.rst", "python/nn/_autosummary_functions/mlx.nn.softmin.rst", "python/nn/_autosummary_functions/mlx.nn.softplus.rst", "python/nn/_autosummary_functions/mlx.nn.softshrink.rst", "python/nn/_autosummary_functions/mlx.nn.step.rst", "python/nn/_autosummary_functions/mlx.nn.tanh.rst", "python/nn/functions.rst", "python/nn/init.rst", "python/nn/layers.rst", "python/nn/losses.rst", "python/nn/module.rst", "python/ops.rst", "python/optimizers.rst", "python/optimizers/_autosummary/mlx.optimizers.AdaDelta.rst", "python/optimizers/_autosummary/mlx.optimizers.Adafactor.rst", "python/optimizers/_autosummary/mlx.optimizers.Adagrad.rst", "python/optimizers/_autosummary/mlx.optimizers.Adam.rst", "python/optimizers/_autosummary/mlx.optimizers.AdamW.rst", "python/optimizers/_autosummary/mlx.optimizers.Adamax.rst", "python/optimizers/_autosummary/mlx.optimizers.Lion.rst", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.apply_gradients.rst", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.init.rst", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.state.rst", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.update.rst", "python/optimizers/_autosummary/mlx.optimizers.RMSprop.rst", "python/optimizers/_autosummary/mlx.optimizers.SGD.rst", "python/optimizers/_autosummary/mlx.optimizers.cosine_decay.rst", "python/optimizers/_autosummary/mlx.optimizers.exponential_decay.rst", "python/optimizers/_autosummary/mlx.optimizers.join_schedules.rst", "python/optimizers/_autosummary/mlx.optimizers.linear_schedule.rst", "python/optimizers/_autosummary/mlx.optimizers.step_decay.rst", "python/optimizers/common_optimizers.rst", "python/optimizers/optimizer.rst", "python/optimizers/schedulers.rst", "python/random.rst", "python/transforms.rst", "python/tree_utils.rst", "usage/compile.rst", "usage/distributed.rst", "usage/function_transforms.rst", "usage/indexing.rst", "usage/lazy_evaluation.rst", "usage/numpy.rst", "usage/quick_start.rst", "usage/saving_and_loading.rst", "usage/unified_memory.rst", "usage/using_streams.rst"], "indexentries": {"__init__() (array method)": [[30, "mlx.core.array.__init__", false]], "__init__() (custom_function method)": [[112, "mlx.core.custom_function.__init__", false]], "__init__() (device method)": [[9, "mlx.core.Device.__init__", false]], "__init__() (dtype method)": [[10, "mlx.core.Dtype.__init__", false]], "__init__() (dtypecategory method)": [[11, "mlx.core.DtypeCategory.__init__", false]], "__init__() (group method)": [[120, "mlx.core.distributed.Group.__init__", false]], "__init__() (stream method)": [[315, "mlx.core.Stream.__init__", false]], "abs (c++ function)": [[0, "_CPPv43absRK5array14StreamOrDevice", false]], "abs() (array method)": [[32, "mlx.core.array.abs", false]], "abs() (in module mlx.core)": [[12, "mlx.core.abs", false]], "adadelta (class in mlx.optimizers)": [[455, "mlx.optimizers.AdaDelta", false]], "adafactor (class in mlx.optimizers)": [[456, "mlx.optimizers.Adafactor", false]], "adagrad (class in mlx.optimizers)": [[457, "mlx.optimizers.Adagrad", false]], "adam (class in mlx.optimizers)": [[458, "mlx.optimizers.Adam", false]], "adamax (class in mlx.optimizers)": [[460, "mlx.optimizers.Adamax", false]], "adamw (class in mlx.optimizers)": [[459, "mlx.optimizers.AdamW", false]], "add (c++ function)": [[0, "_CPPv43addRK5arrayRK5array14StreamOrDevice", false]], "add() (in module mlx.core)": [[13, "mlx.core.add", false]], "addmm (c++ function)": [[0, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", false]], "addmm() (in module mlx.core)": [[14, "mlx.core.addmm", false]], "affine_quantize() (in module mlx.core.fast)": [[141, "mlx.core.fast.affine_quantize", false]], "alibi (class in mlx.nn)": [[325, "mlx.nn.ALiBi", false]], "all (c++ function)": [[0, "_CPPv43allRK5array14StreamOrDevice", false], [0, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv43allRK5arrayb14StreamOrDevice", false], [0, "_CPPv43allRK5arrayib14StreamOrDevice", false]], "all() (array method)": [[33, "mlx.core.array.all", false]], "all() (in module mlx.core)": [[15, "mlx.core.all", false]], "all_gather() (in module mlx.core.distributed)": [[121, "mlx.core.distributed.all_gather", false]], "all_sum() (in module mlx.core.distributed)": [[122, "mlx.core.distributed.all_sum", false]], "allclose (c++ function)": [[0, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", false]], "allclose() (in module mlx.core)": [[16, "mlx.core.allclose", false]], "any (c++ function)": [[0, "_CPPv43anyRK5array14StreamOrDevice", false], [0, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv43anyRK5arrayb14StreamOrDevice", false], [0, "_CPPv43anyRK5arrayib14StreamOrDevice", false]], "any() (array method)": [[34, "mlx.core.array.any", false]], "any() (in module mlx.core)": [[17, "mlx.core.any", false]], "apply() (module method)": [[358, "mlx.nn.Module.apply", false]], "apply_gradients() (optimizer method)": [[462, "mlx.optimizers.Optimizer.apply_gradients", false]], "apply_to_modules() (module method)": [[359, "mlx.nn.Module.apply_to_modules", false]], "arange (c++ function)": [[0, "_CPPv46aranged14StreamOrDevice", false], [0, "_CPPv46aranged5Dtype14StreamOrDevice", false], [0, "_CPPv46arangedd14StreamOrDevice", false], [0, "_CPPv46arangedd5Dtype14StreamOrDevice", false], [0, "_CPPv46arangeddd14StreamOrDevice", false], [0, "_CPPv46arangeddd5Dtype14StreamOrDevice", false], [0, "_CPPv46arangei14StreamOrDevice", false], [0, "_CPPv46arangeii14StreamOrDevice", false], [0, "_CPPv46arangeiii14StreamOrDevice", false]], "arange() (in module mlx.core)": [[18, "mlx.core.arange", false]], "arccos (c++ function)": [[0, "_CPPv46arccosRK5array14StreamOrDevice", false]], "arccos() (in module mlx.core)": [[19, "mlx.core.arccos", false]], "arccosh (c++ function)": [[0, "_CPPv47arccoshRK5array14StreamOrDevice", false]], "arccosh() (in module mlx.core)": [[20, "mlx.core.arccosh", false]], "arcsin (c++ function)": [[0, "_CPPv46arcsinRK5array14StreamOrDevice", false]], "arcsin() (in module mlx.core)": [[21, "mlx.core.arcsin", false]], "arcsinh (c++ function)": [[0, "_CPPv47arcsinhRK5array14StreamOrDevice", false]], "arcsinh() (in module mlx.core)": [[22, "mlx.core.arcsinh", false]], "arctan (c++ function)": [[0, "_CPPv46arctanRK5array14StreamOrDevice", false]], "arctan() (in module mlx.core)": [[23, "mlx.core.arctan", false]], "arctan2 (c++ function)": [[0, "_CPPv47arctan2RK5arrayRK5array14StreamOrDevice", false]], "arctan2() (in module mlx.core)": [[24, "mlx.core.arctan2", false]], "arctanh (c++ function)": [[0, "_CPPv47arctanhRK5array14StreamOrDevice", false]], "arctanh() (in module mlx.core)": [[25, "mlx.core.arctanh", false]], "argmax (c++ function)": [[0, "_CPPv46argmaxRK5array14StreamOrDevice", false], [0, "_CPPv46argmaxRK5arrayb14StreamOrDevice", false], [0, "_CPPv46argmaxRK5arrayib14StreamOrDevice", false]], "argmax() (array method)": [[35, "mlx.core.array.argmax", false]], "argmax() (in module mlx.core)": [[26, "mlx.core.argmax", false]], "argmin (c++ function)": [[0, "_CPPv46argminRK5array14StreamOrDevice", false], [0, "_CPPv46argminRK5arrayb14StreamOrDevice", false], [0, "_CPPv46argminRK5arrayib14StreamOrDevice", false]], "argmin() (array method)": [[36, "mlx.core.array.argmin", false]], "argmin() (in module mlx.core)": [[27, "mlx.core.argmin", false]], "argpartition (c++ function)": [[0, "_CPPv412argpartitionRK5arrayi14StreamOrDevice", false], [0, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", false]], "argpartition() (in module mlx.core)": [[28, "mlx.core.argpartition", false]], "argsort (c++ function)": [[0, "_CPPv47argsortRK5array14StreamOrDevice", false], [0, "_CPPv47argsortRK5arrayi14StreamOrDevice", false]], "argsort() (in module mlx.core)": [[29, "mlx.core.argsort", false]], "array (class in mlx.core)": [[30, "mlx.core.array", false]], "array_equal (c++ function)": [[0, "_CPPv411array_equalRK5arrayRK5array14StreamOrDevice", false], [0, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", false]], "array_equal() (in module mlx.core)": [[82, "mlx.core.array_equal", false]], "as_strided (c++ function)": [[0, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", false]], "as_strided() (in module mlx.core)": [[83, "mlx.core.as_strided", false]], "astype (c++ function)": [[0, "_CPPv46astype5array5Dtype14StreamOrDevice", false]], "astype() (array method)": [[37, "mlx.core.array.astype", false]], "at (array property)": [[38, "mlx.core.array.at", false]], "atleast_1d (c++ function)": [[0, "_CPPv410atleast_1dRK5array14StreamOrDevice", false], [0, "_CPPv410atleast_1dRKNSt6vectorI5arrayEE14StreamOrDevice", false]], "atleast_1d() (in module mlx.core)": [[84, "mlx.core.atleast_1d", false]], "atleast_2d (c++ function)": [[0, "_CPPv410atleast_2dRK5array14StreamOrDevice", false], [0, "_CPPv410atleast_2dRKNSt6vectorI5arrayEE14StreamOrDevice", false]], "atleast_2d() (in module mlx.core)": [[85, "mlx.core.atleast_2d", false]], "atleast_3d (c++ function)": [[0, "_CPPv410atleast_3dRK5array14StreamOrDevice", false], [0, "_CPPv410atleast_3dRKNSt6vectorI5arrayEE14StreamOrDevice", false]], "atleast_3d() (in module mlx.core)": [[86, "mlx.core.atleast_3d", false]], "avgpool1d (class in mlx.nn)": [[326, "mlx.nn.AvgPool1d", false]], "avgpool2d (class in mlx.nn)": [[327, "mlx.nn.AvgPool2d", false]], "batchnorm (class in mlx.nn)": [[328, "mlx.nn.BatchNorm", false]], "bernoulli() (in module mlx.core.random)": [[240, "mlx.core.random.bernoulli", false]], "binary_cross_entropy (class in mlx.nn.losses)": [[421, "mlx.nn.losses.binary_cross_entropy", false]], "bitwise_and (c++ function)": [[0, "_CPPv411bitwise_andRK5arrayRK5array14StreamOrDevice", false]], "bitwise_and() (in module mlx.core)": [[87, "mlx.core.bitwise_and", false]], "bitwise_or (c++ function)": [[0, "_CPPv410bitwise_orRK5arrayRK5array14StreamOrDevice", false]], "bitwise_or() (in module mlx.core)": [[88, "mlx.core.bitwise_or", false]], "bitwise_xor (c++ function)": [[0, "_CPPv411bitwise_xorRK5arrayRK5array14StreamOrDevice", false]], "bitwise_xor() (in module mlx.core)": [[89, "mlx.core.bitwise_xor", false]], "block_masked_mm (c++ function)": [[0, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", false]], "block_masked_mm() (in module mlx.core)": [[90, "mlx.core.block_masked_mm", false]], "broadcast_arrays (c++ function)": [[0, "_CPPv416broadcast_arraysRKNSt6vectorI5arrayEE14StreamOrDevice", false]], "broadcast_to (c++ function)": [[0, "_CPPv412broadcast_toRK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "broadcast_to() (in module mlx.core)": [[91, "mlx.core.broadcast_to", false]], "categorical() (in module mlx.core.random)": [[241, "mlx.core.random.categorical", false]], "ceil (c++ function)": [[0, "_CPPv44ceilRK5array14StreamOrDevice", false]], "ceil() (in module mlx.core)": [[92, "mlx.core.ceil", false]], "celu (class in mlx.nn)": [[329, "mlx.nn.CELU", false], [409, "mlx.nn.celu", false]], "children() (module method)": [[360, "mlx.nn.Module.children", false]], "cholesky() (in module mlx.core.linalg)": [[183, "mlx.core.linalg.cholesky", false]], "cholesky_inv() (in module mlx.core.linalg)": [[184, "mlx.core.linalg.cholesky_inv", false]], "clear_cache() (in module mlx.core.metal)": [[209, "mlx.core.metal.clear_cache", false]], "clip (c++ function)": [[0, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", false]], "clip() (in module mlx.core)": [[93, "mlx.core.clip", false]], "clip_grad_norm() (in module mlx.optimizers)": [[309, "mlx.optimizers.clip_grad_norm", false]], "compile() (in module mlx.core)": [[94, "mlx.core.compile", false]], "concatenate (c++ function)": [[0, "_CPPv411concatenateRKNSt6vectorI5arrayEE14StreamOrDevice", false], [0, "_CPPv411concatenateRKNSt6vectorI5arrayEEi14StreamOrDevice", false]], "concatenate() (in module mlx.core)": [[95, "mlx.core.concatenate", false]], "conj() (array method)": [[39, "mlx.core.array.conj", false]], "conj() (in module mlx.core)": [[96, "mlx.core.conj", false]], "conjugate (c++ function)": [[0, "_CPPv49conjugateRK5array14StreamOrDevice", false]], "conjugate() (in module mlx.core)": [[97, "mlx.core.conjugate", false]], "constant() (in module mlx.nn.init)": [[401, "mlx.nn.init.constant", false]], "conv1d (c++ function)": [[0, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", false]], "conv1d (class in mlx.nn)": [[330, "mlx.nn.Conv1d", false]], "conv1d() (in module mlx.core)": [[98, "mlx.core.conv1d", false]], "conv2d (c++ function)": [[0, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", false]], "conv2d (class in mlx.nn)": [[331, "mlx.nn.Conv2d", false]], "conv2d() (in module mlx.core)": [[99, "mlx.core.conv2d", false]], "conv3d (c++ function)": [[0, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", false]], "conv3d (class in mlx.nn)": [[332, "mlx.nn.Conv3d", false]], "conv3d() (in module mlx.core)": [[100, "mlx.core.conv3d", false]], "conv_general (c++ function)": [[0, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", false], [0, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", false]], "conv_general() (in module mlx.core)": [[101, "mlx.core.conv_general", false]], "conv_transpose1d (c++ function)": [[0, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", false]], "conv_transpose1d() (in module mlx.core)": [[102, "mlx.core.conv_transpose1d", false]], "conv_transpose2d (c++ function)": [[0, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", false]], "conv_transpose2d() (in module mlx.core)": [[103, "mlx.core.conv_transpose2d", false]], "conv_transpose3d (c++ function)": [[0, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", false]], "conv_transpose3d() (in module mlx.core)": [[104, "mlx.core.conv_transpose3d", false]], "convolve() (in module mlx.core)": [[105, "mlx.core.convolve", false]], "convtranspose1d (class in mlx.nn)": [[333, "mlx.nn.ConvTranspose1d", false]], "convtranspose2d (class in mlx.nn)": [[334, "mlx.nn.ConvTranspose2d", false]], "convtranspose3d (class in mlx.nn)": [[335, "mlx.nn.ConvTranspose3d", false]], "copy (c++ function)": [[0, "_CPPv44copy5array14StreamOrDevice", false]], "cos (c++ function)": [[0, "_CPPv43cosRK5array14StreamOrDevice", false]], "cos() (array method)": [[40, "mlx.core.array.cos", false]], "cos() (in module mlx.core)": [[106, "mlx.core.cos", false]], "cosh (c++ function)": [[0, "_CPPv44coshRK5array14StreamOrDevice", false]], "cosh() (in module mlx.core)": [[107, "mlx.core.cosh", false]], "cosine_decay() (in module mlx.optimizers)": [[468, "mlx.optimizers.cosine_decay", false]], "cosine_similarity_loss (class in mlx.nn.losses)": [[422, "mlx.nn.losses.cosine_similarity_loss", false]], "cross() (in module mlx.core.linalg)": [[185, "mlx.core.linalg.cross", false]], "cross_entropy (class in mlx.nn.losses)": [[423, "mlx.nn.losses.cross_entropy", false]], "cummax (c++ function)": [[0, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", false]], "cummax() (array method)": [[41, "mlx.core.array.cummax", false]], "cummax() (in module mlx.core)": [[108, "mlx.core.cummax", false]], "cummin (c++ function)": [[0, "_CPPv46cumminRK5arrayibb14StreamOrDevice", false]], "cummin() (array method)": [[42, "mlx.core.array.cummin", false]], "cummin() (in module mlx.core)": [[109, "mlx.core.cummin", false]], "cumprod (c++ function)": [[0, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", false]], "cumprod() (array method)": [[43, "mlx.core.array.cumprod", false]], "cumprod() (in module mlx.core)": [[110, "mlx.core.cumprod", false]], "cumsum (c++ function)": [[0, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", false]], "cumsum() (array method)": [[44, "mlx.core.array.cumsum", false]], "cumsum() (in module mlx.core)": [[111, "mlx.core.cumsum", false]], "custom_function (class in mlx.core)": [[112, "mlx.core.custom_function", false]], "default_device() (in module mlx.core)": [[113, "mlx.core.default_device", false]], "default_stream() (in module mlx.core)": [[114, "mlx.core.default_stream", false]], "degrees (c++ function)": [[0, "_CPPv47degreesRK5array14StreamOrDevice", false]], "degrees() (in module mlx.core)": [[115, "mlx.core.degrees", false]], "depends (c++ function)": [[0, "_CPPv47dependsRKNSt6vectorI5arrayEERKNSt6vectorI5arrayEE", false]], "dequantize (c++ function)": [[0, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", false]], "dequantize() (in module mlx.core)": [[116, "mlx.core.dequantize", false]], "device (class in mlx.core)": [[9, "mlx.core.Device", false]], "device_info() (in module mlx.core.metal)": [[210, "mlx.core.metal.device_info", false]], "diag (c++ function)": [[0, "_CPPv44diagRK5arrayi14StreamOrDevice", false]], "diag() (array method)": [[45, "mlx.core.array.diag", false]], "diag() (in module mlx.core)": [[117, "mlx.core.diag", false]], "diagonal (c++ function)": [[0, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", false]], "diagonal() (array method)": [[46, "mlx.core.array.diagonal", false]], "diagonal() (in module mlx.core)": [[118, "mlx.core.diagonal", false]], "disable_compile() (in module mlx.core)": [[119, "mlx.core.disable_compile", false]], "divide (c++ function)": [[0, "_CPPv46divideRK5arrayRK5array14StreamOrDevice", false]], "divide() (in module mlx.core)": [[128, "mlx.core.divide", false]], "divmod (c++ function)": [[0, "_CPPv46divmodRK5arrayRK5array14StreamOrDevice", false]], "divmod() (in module mlx.core)": [[129, "mlx.core.divmod", false]], "dropout (class in mlx.nn)": [[336, "mlx.nn.Dropout", false]], "dropout2d (class in mlx.nn)": [[337, "mlx.nn.Dropout2d", false]], "dropout3d (class in mlx.nn)": [[338, "mlx.nn.Dropout3d", false]], "dtype (array property)": [[47, "mlx.core.array.dtype", false]], "dtype (class in mlx.core)": [[10, "mlx.core.Dtype", false]], "dtypecategory (class in mlx.core)": [[11, "mlx.core.DtypeCategory", false]], "eigh() (in module mlx.core.linalg)": [[186, "mlx.core.linalg.eigh", false]], "eigvalsh() (in module mlx.core.linalg)": [[187, "mlx.core.linalg.eigvalsh", false]], "einsum() (in module mlx.core)": [[130, "mlx.core.einsum", false]], "einsum_path() (in module mlx.core)": [[131, "mlx.core.einsum_path", false]], "elu (class in mlx.nn)": [[339, "mlx.nn.ELU", false], [410, "mlx.nn.elu", false]], "embedding (class in mlx.nn)": [[340, "mlx.nn.Embedding", false]], "enable_compile() (in module mlx.core)": [[132, "mlx.core.enable_compile", false]], "equal (c++ function)": [[0, "_CPPv45equalRK5arrayRK5array14StreamOrDevice", false]], "equal() (in module mlx.core)": [[133, "mlx.core.equal", false]], "erf (c++ function)": [[0, "_CPPv43erfRK5array14StreamOrDevice", false]], "erf() (in module mlx.core)": [[134, "mlx.core.erf", false]], "erfinv (c++ function)": [[0, "_CPPv46erfinvRK5array14StreamOrDevice", false]], "erfinv() (in module mlx.core)": [[135, "mlx.core.erfinv", false]], "eval() (in module mlx.core)": [[136, "mlx.core.eval", false]], "eval() (module method)": [[361, "mlx.nn.Module.eval", false]], "exp (c++ function)": [[0, "_CPPv43expRK5array14StreamOrDevice", false]], "exp() (array method)": [[48, "mlx.core.array.exp", false]], "exp() (in module mlx.core)": [[137, "mlx.core.exp", false]], "expand_dims (c++ function)": [[0, "_CPPv411expand_dimsRK5arrayRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv411expand_dimsRK5arrayi14StreamOrDevice", false]], "expand_dims() (in module mlx.core)": [[138, "mlx.core.expand_dims", false]], "expm1 (c++ function)": [[0, "_CPPv45expm1RK5array14StreamOrDevice", false]], "expm1() (in module mlx.core)": [[139, "mlx.core.expm1", false]], "exponential_decay() (in module mlx.optimizers)": [[469, "mlx.optimizers.exponential_decay", false]], "eye (c++ function)": [[0, "_CPPv43eyei14StreamOrDevice", false], [0, "_CPPv43eyei5Dtype14StreamOrDevice", false], [0, "_CPPv43eyeii14StreamOrDevice", false], [0, "_CPPv43eyeiii14StreamOrDevice", false], [0, "_CPPv43eyeiii5Dtype14StreamOrDevice", false]], "eye() (in module mlx.core)": [[140, "mlx.core.eye", false]], "fft() (in module mlx.core.fft)": [[147, "mlx.core.fft.fft", false]], "fft2() (in module mlx.core.fft)": [[148, "mlx.core.fft.fft2", false]], "fftn() (in module mlx.core.fft)": [[149, "mlx.core.fft.fftn", false]], "filter_and_map() (module method)": [[362, "mlx.nn.Module.filter_and_map", false]], "flatten (c++ function)": [[0, "_CPPv47flattenRK5array14StreamOrDevice", false], [0, "_CPPv47flattenRK5arrayii14StreamOrDevice", false]], "flatten() (array method)": [[49, "mlx.core.array.flatten", false]], "flatten() (in module mlx.core)": [[159, "mlx.core.flatten", false]], "floor (c++ function)": [[0, "_CPPv45floorRK5array14StreamOrDevice", false]], "floor() (in module mlx.core)": [[160, "mlx.core.floor", false]], "floor_divide (c++ function)": [[0, "_CPPv412floor_divideRK5arrayRK5array14StreamOrDevice", false]], "floor_divide() (in module mlx.core)": [[161, "mlx.core.floor_divide", false]], "freeze() (module method)": [[363, "mlx.nn.Module.freeze", false]], "full (c++ function)": [[0, "_CPPv44fullNSt6vectorIiEE5array14StreamOrDevice", false], [0, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", false], [0, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", false], [0, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", false]], "full() (in module mlx.core)": [[162, "mlx.core.full", false]], "gather (c++ function)": [[0, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", false]], "gather_mm (c++ function)": [[0, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", false]], "gather_mm() (in module mlx.core)": [[163, "mlx.core.gather_mm", false]], "gather_qmm (c++ function)": [[0, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", false]], "gather_qmm() (in module mlx.core)": [[164, "mlx.core.gather_qmm", false]], "gaussian_nll_loss (class in mlx.nn.losses)": [[424, "mlx.nn.losses.gaussian_nll_loss", false]], "gelu (class in mlx.nn)": [[341, "mlx.nn.GELU", false], [411, "mlx.nn.gelu", false]], "gelu_approx (class in mlx.nn)": [[412, "mlx.nn.gelu_approx", false]], "gelu_fast_approx (class in mlx.nn)": [[413, "mlx.nn.gelu_fast_approx", false]], "get_active_memory() (in module mlx.core.metal)": [[211, "mlx.core.metal.get_active_memory", false]], "get_cache_memory() (in module mlx.core.metal)": [[212, "mlx.core.metal.get_cache_memory", false]], "get_peak_memory() (in module mlx.core.metal)": [[213, "mlx.core.metal.get_peak_memory", false]], "glorot_normal() (in module mlx.nn.init)": [[402, "mlx.nn.init.glorot_normal", false]], "glorot_uniform() (in module mlx.nn.init)": [[403, "mlx.nn.init.glorot_uniform", false]], "glu (class in mlx.nn)": [[342, "mlx.nn.GLU", false], [414, "mlx.nn.glu", false]], "grad() (in module mlx.core)": [[165, "mlx.core.grad", false]], "greater (c++ function)": [[0, "_CPPv47greaterRK5arrayRK5array14StreamOrDevice", false]], "greater() (in module mlx.core)": [[166, "mlx.core.greater", false]], "greater_equal (c++ function)": [[0, "_CPPv413greater_equalRK5arrayRK5array14StreamOrDevice", false]], "greater_equal() (in module mlx.core)": [[167, "mlx.core.greater_equal", false]], "group (class in mlx.core.distributed)": [[120, "mlx.core.distributed.Group", false]], "groupnorm (class in mlx.nn)": [[344, "mlx.nn.GroupNorm", false]], "gru (class in mlx.nn)": [[343, "mlx.nn.GRU", false]], "gumbel() (in module mlx.core.random)": [[242, "mlx.core.random.gumbel", false]], "hadamard_transform (c++ function)": [[0, "_CPPv418hadamard_transformRK5arrayNSt8optionalIfEE14StreamOrDevice", false]], "hadamard_transform() (in module mlx.core)": [[168, "mlx.core.hadamard_transform", false]], "hard_shrink (class in mlx.nn)": [[415, "mlx.nn.hard_shrink", false]], "hard_tanh (class in mlx.nn)": [[416, "mlx.nn.hard_tanh", false]], "hardshrink (class in mlx.nn)": [[345, "mlx.nn.HardShrink", false]], "hardswish (class in mlx.nn)": [[347, "mlx.nn.Hardswish", false], [417, "mlx.nn.hardswish", false]], "hardtanh (class in mlx.nn)": [[346, "mlx.nn.HardTanh", false]], "he_normal() (in module mlx.nn.init)": [[404, "mlx.nn.init.he_normal", false]], "he_uniform() (in module mlx.nn.init)": [[405, "mlx.nn.init.he_uniform", false]], "hinge_loss (class in mlx.nn.losses)": [[425, "mlx.nn.losses.hinge_loss", false]], "huber_loss (class in mlx.nn.losses)": [[426, "mlx.nn.losses.huber_loss", false]], "identity (c++ function)": [[0, "_CPPv48identityi14StreamOrDevice", false], [0, "_CPPv48identityi5Dtype14StreamOrDevice", false]], "identity() (in module mlx.core)": [[169, "mlx.core.identity", false]], "identity() (in module mlx.nn.init)": [[406, "mlx.nn.init.identity", false]], "ifft() (in module mlx.core.fft)": [[150, "mlx.core.fft.ifft", false]], "ifft2() (in module mlx.core.fft)": [[151, "mlx.core.fft.ifft2", false]], "ifftn() (in module mlx.core.fft)": [[152, "mlx.core.fft.ifftn", false]], "imag (c++ function)": [[0, "_CPPv44imagRK5array14StreamOrDevice", false]], "imag() (in module mlx.core)": [[170, "mlx.core.imag", false]], "init() (in module mlx.core.distributed)": [[123, "mlx.core.distributed.init", false]], "init() (optimizer method)": [[463, "mlx.optimizers.Optimizer.init", false]], "inner (c++ function)": [[0, "_CPPv45innerRK5arrayRK5array14StreamOrDevice", false]], "inner() (in module mlx.core)": [[171, "mlx.core.inner", false]], "instancenorm (class in mlx.nn)": [[348, "mlx.nn.InstanceNorm", false]], "inv() (in module mlx.core.linalg)": [[188, "mlx.core.linalg.inv", false]], "irfft() (in module mlx.core.fft)": [[153, "mlx.core.fft.irfft", false]], "irfft2() (in module mlx.core.fft)": [[154, "mlx.core.fft.irfft2", false]], "irfftn() (in module mlx.core.fft)": [[155, "mlx.core.fft.irfftn", false]], "is_available() (in module mlx.core.distributed)": [[124, "mlx.core.distributed.is_available", false]], "is_available() (in module mlx.core.metal)": [[214, "mlx.core.metal.is_available", false]], "isclose (c++ function)": [[0, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", false]], "isclose() (in module mlx.core)": [[172, "mlx.core.isclose", false]], "isfinite (c++ function)": [[0, "_CPPv48isfiniteRK5array14StreamOrDevice", false]], "isfinite() (in module mlx.core)": [[173, "mlx.core.isfinite", false]], "isinf (c++ function)": [[0, "_CPPv45isinfRK5array14StreamOrDevice", false]], "isinf() (in module mlx.core)": [[174, "mlx.core.isinf", false]], "isnan (c++ function)": [[0, "_CPPv45isnanRK5array14StreamOrDevice", false]], "isnan() (in module mlx.core)": [[175, "mlx.core.isnan", false]], "isneginf (c++ function)": [[0, "_CPPv48isneginfRK5array14StreamOrDevice", false]], "isneginf() (in module mlx.core)": [[176, "mlx.core.isneginf", false]], "isposinf (c++ function)": [[0, "_CPPv48isposinfRK5array14StreamOrDevice", false]], "isposinf() (in module mlx.core)": [[177, "mlx.core.isposinf", false]], "issubdtype() (in module mlx.core)": [[178, "mlx.core.issubdtype", false]], "item() (array method)": [[50, "mlx.core.array.item", false]], "itemsize (array property)": [[51, "mlx.core.array.itemsize", false]], "join_schedules() (in module mlx.optimizers)": [[470, "mlx.optimizers.join_schedules", false]], "jvp() (in module mlx.core)": [[179, "mlx.core.jvp", false]], "key() (in module mlx.core.random)": [[243, "mlx.core.random.key", false]], "kl_div_loss (class in mlx.nn.losses)": [[427, "mlx.nn.losses.kl_div_loss", false]], "l1_loss (class in mlx.nn.losses)": [[428, "mlx.nn.losses.l1_loss", false]], "laplace() (in module mlx.core.random)": [[244, "mlx.core.random.laplace", false]], "layer_norm() (in module mlx.core.fast)": [[142, "mlx.core.fast.layer_norm", false]], "layernorm (class in mlx.nn)": [[350, "mlx.nn.LayerNorm", false]], "leaf_modules() (module method)": [[364, "mlx.nn.Module.leaf_modules", false]], "leaky_relu (class in mlx.nn)": [[418, "mlx.nn.leaky_relu", false]], "leakyrelu (class in mlx.nn)": [[351, "mlx.nn.LeakyReLU", false]], "left_shift (c++ function)": [[0, "_CPPv410left_shiftRK5arrayRK5array14StreamOrDevice", false]], "left_shift() (in module mlx.core)": [[180, "mlx.core.left_shift", false]], "less (c++ function)": [[0, "_CPPv44lessRK5arrayRK5array14StreamOrDevice", false]], "less() (in module mlx.core)": [[181, "mlx.core.less", false]], "less_equal (c++ function)": [[0, "_CPPv410less_equalRK5arrayRK5array14StreamOrDevice", false]], "less_equal() (in module mlx.core)": [[182, "mlx.core.less_equal", false]], "linear (class in mlx.nn)": [[352, "mlx.nn.Linear", false]], "linear_schedule() (in module mlx.optimizers)": [[471, "mlx.optimizers.linear_schedule", false]], "linspace (c++ function)": [[0, "_CPPv48linspaceddi5Dtype14StreamOrDevice", false]], "linspace() (in module mlx.core)": [[193, "mlx.core.linspace", false]], "lion (class in mlx.optimizers)": [[461, "mlx.optimizers.Lion", false]], "load() (in module mlx.core)": [[194, "mlx.core.load", false]], "load_weights() (module method)": [[365, "mlx.nn.Module.load_weights", false]], "log (c++ function)": [[0, "_CPPv43logRK5array14StreamOrDevice", false]], "log() (array method)": [[52, "mlx.core.array.log", false]], "log() (in module mlx.core)": [[195, "mlx.core.log", false]], "log10 (c++ function)": [[0, "_CPPv45log10RK5array14StreamOrDevice", false]], "log10() (array method)": [[53, "mlx.core.array.log10", false]], "log10() (in module mlx.core)": [[196, "mlx.core.log10", false]], "log1p (c++ function)": [[0, "_CPPv45log1pRK5array14StreamOrDevice", false]], "log1p() (array method)": [[54, "mlx.core.array.log1p", false]], "log1p() (in module mlx.core)": [[197, "mlx.core.log1p", false]], "log2 (c++ function)": [[0, "_CPPv44log2RK5array14StreamOrDevice", false]], "log2() (array method)": [[55, "mlx.core.array.log2", false]], "log2() (in module mlx.core)": [[198, "mlx.core.log2", false]], "log_cosh_loss (class in mlx.nn.losses)": [[429, "mlx.nn.losses.log_cosh_loss", false]], "log_sigmoid (class in mlx.nn)": [[419, "mlx.nn.log_sigmoid", false]], "log_softmax (class in mlx.nn)": [[420, "mlx.nn.log_softmax", false]], "logaddexp (c++ function)": [[0, "_CPPv49logaddexpRK5arrayRK5array14StreamOrDevice", false]], "logaddexp() (in module mlx.core)": [[199, "mlx.core.logaddexp", false]], "logical_and (c++ function)": [[0, "_CPPv411logical_andRK5arrayRK5array14StreamOrDevice", false]], "logical_and() (in module mlx.core)": [[200, "mlx.core.logical_and", false]], "logical_not (c++ function)": [[0, "_CPPv411logical_notRK5array14StreamOrDevice", false]], "logical_not() (in module mlx.core)": [[201, "mlx.core.logical_not", false]], "logical_or (c++ function)": [[0, "_CPPv410logical_orRK5arrayRK5array14StreamOrDevice", false]], "logical_or() (in module mlx.core)": [[202, "mlx.core.logical_or", false]], "logsigmoid (class in mlx.nn)": [[353, "mlx.nn.LogSigmoid", false]], "logsoftmax (class in mlx.nn)": [[354, "mlx.nn.LogSoftmax", false]], "logsumexp (c++ function)": [[0, "_CPPv49logsumexpRK5array14StreamOrDevice", false], [0, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv49logsumexpRK5arrayb14StreamOrDevice", false], [0, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", false]], "logsumexp() (array method)": [[56, "mlx.core.array.logsumexp", false]], "logsumexp() (in module mlx.core)": [[203, "mlx.core.logsumexp", false]], "lstm (class in mlx.nn)": [[349, "mlx.nn.LSTM", false]], "margin_ranking_loss (class in mlx.nn.losses)": [[430, "mlx.nn.losses.margin_ranking_loss", false]], "matmul (c++ function)": [[0, "_CPPv46matmulRK5arrayRK5array14StreamOrDevice", false]], "matmul() (in module mlx.core)": [[204, "mlx.core.matmul", false]], "max (c++ function)": [[0, "_CPPv43maxRK5array14StreamOrDevice", false], [0, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv43maxRK5arrayb14StreamOrDevice", false], [0, "_CPPv43maxRK5arrayib14StreamOrDevice", false]], "max() (array method)": [[57, "mlx.core.array.max", false]], "max() (in module mlx.core)": [[205, "mlx.core.max", false]], "maximum (c++ function)": [[0, "_CPPv47maximumRK5arrayRK5array14StreamOrDevice", false]], "maximum() (in module mlx.core)": [[206, "mlx.core.maximum", false]], "maxpool1d (class in mlx.nn)": [[355, "mlx.nn.MaxPool1d", false]], "maxpool2d (class in mlx.nn)": [[356, "mlx.nn.MaxPool2d", false]], "mean (c++ function)": [[0, "_CPPv44meanRK5array14StreamOrDevice", false], [0, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv44meanRK5arrayb14StreamOrDevice", false], [0, "_CPPv44meanRK5arrayib14StreamOrDevice", false]], "mean() (array method)": [[58, "mlx.core.array.mean", false]], "mean() (in module mlx.core)": [[207, "mlx.core.mean", false]], "meshgrid (c++ function)": [[0, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", false]], "meshgrid() (in module mlx.core)": [[208, "mlx.core.meshgrid", false]], "metal_kernel() (in module mlx.core.fast)": [[143, "mlx.core.fast.metal_kernel", false]], "min (c++ function)": [[0, "_CPPv43minRK5array14StreamOrDevice", false], [0, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv43minRK5arrayb14StreamOrDevice", false], [0, "_CPPv43minRK5arrayib14StreamOrDevice", false]], "min() (array method)": [[59, "mlx.core.array.min", false]], "min() (in module mlx.core)": [[221, "mlx.core.min", false]], "minimum (c++ function)": [[0, "_CPPv47minimumRK5arrayRK5array14StreamOrDevice", false]], "minimum() (in module mlx.core)": [[222, "mlx.core.minimum", false]], "mish (class in mlx.nn)": [[357, "mlx.nn.Mish", false], [435, "mlx.nn.mish", false]], "module (class in mlx.nn)": [[452, "mlx.nn.Module", false]], "modules() (module method)": [[366, "mlx.nn.Module.modules", false]], "moveaxis (c++ function)": [[0, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", false]], "moveaxis() (array method)": [[60, "mlx.core.array.moveaxis", false]], "moveaxis() (in module mlx.core)": [[223, "mlx.core.moveaxis", false]], "mse_loss (class in mlx.nn.losses)": [[431, "mlx.nn.losses.mse_loss", false]], "multiheadattention (class in mlx.nn)": [[378, "mlx.nn.MultiHeadAttention", false]], "multiply (c++ function)": [[0, "_CPPv48multiplyRK5arrayRK5array14StreamOrDevice", false]], "multiply() (in module mlx.core)": [[224, "mlx.core.multiply", false]], "multivariate_normal() (in module mlx.core.random)": [[245, "mlx.core.random.multivariate_normal", false]], "named_modules() (module method)": [[367, "mlx.nn.Module.named_modules", false]], "nan_to_num (c++ function)": [[0, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", false]], "nan_to_num() (in module mlx.core)": [[225, "mlx.core.nan_to_num", false]], "nbytes (array property)": [[61, "mlx.core.array.nbytes", false]], "ndim (array property)": [[62, "mlx.core.array.ndim", false]], "negative (c++ function)": [[0, "_CPPv48negativeRK5array14StreamOrDevice", false]], "negative() (in module mlx.core)": [[226, "mlx.core.negative", false]], "new_stream() (in module mlx.core)": [[227, "mlx.core.new_stream", false]], "nll_loss (class in mlx.nn.losses)": [[432, "mlx.nn.losses.nll_loss", false]], "norm() (in module mlx.core.linalg)": [[189, "mlx.core.linalg.norm", false]], "normal() (in module mlx.core.random)": [[246, "mlx.core.random.normal", false]], "normal() (in module mlx.nn.init)": [[407, "mlx.nn.init.normal", false]], "not_equal (c++ function)": [[0, "_CPPv49not_equalRK5arrayRK5array14StreamOrDevice", false]], "not_equal() (in module mlx.core)": [[228, "mlx.core.not_equal", false]], "number_of_elements (c++ function)": [[0, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", false]], "ones (c++ function)": [[0, "_CPPv44onesRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv44onesRKNSt6vectorIiEE5Dtype14StreamOrDevice", false]], "ones() (in module mlx.core)": [[229, "mlx.core.ones", false]], "ones_like (c++ function)": [[0, "_CPPv49ones_likeRK5array14StreamOrDevice", false]], "ones_like() (in module mlx.core)": [[230, "mlx.core.ones_like", false]], "operator!= (c++ function)": [[0, "_CPPv4I0Ene5array1TRK5array", false], [0, "_CPPv4I0Ene5arrayRK5array1T", false], [0, "_CPPv4neRK5arrayRK5array", false]], "operator% (c++ function)": [[0, "_CPPv4I0Erm5array1TRK5array", false], [0, "_CPPv4I0Erm5arrayRK5array1T", false], [0, "_CPPv4rmRK5arrayRK5array", false]], "operator& (c++ function)": [[0, "_CPPv4anRK5arrayRK5array", false]], "operator&& (c++ function)": [[0, "_CPPv4aaRK5arrayRK5array", false]], "operator* (c++ function)": [[0, "_CPPv4I0Eml5array1TRK5array", false], [0, "_CPPv4I0Eml5arrayRK5array1T", false], [0, "_CPPv4mlRK5arrayRK5array", false]], "operator+ (c++ function)": [[0, "_CPPv4I0Epl5array1TRK5array", false], [0, "_CPPv4I0Epl5arrayRK5array1T", false], [0, "_CPPv4plRK5arrayRK5array", false]], "operator- (c++ function)": [[0, "_CPPv4I0Emi5array1TRK5array", false], [0, "_CPPv4I0Emi5arrayRK5array1T", false], [0, "_CPPv4miRK5array", false], [0, "_CPPv4miRK5arrayRK5array", false]], "operator/ (c++ function)": [[0, "_CPPv4dvRK5arrayRK5array", false], [0, "_CPPv4dvRK5arrayd", false], [0, "_CPPv4dvdRK5array", false]], "operator< (c++ function)": [[0, "_CPPv4I0Elt5array1TRK5array", false], [0, "_CPPv4I0Elt5arrayRK5array1T", false], [0, "_CPPv4ltRK5arrayRK5array", false]], "operator<< (c++ function)": [[0, "_CPPv4lsRK5arrayRK5array", false]], "operator<= (c++ function)": [[0, "_CPPv4I0Ele5array1TRK5array", false], [0, "_CPPv4I0Ele5arrayRK5array1T", false], [0, "_CPPv4leRK5arrayRK5array", false]], "operator== (c++ function)": [[0, "_CPPv4I0Eeq5array1TRK5array", false], [0, "_CPPv4I0Eeq5arrayRK5array1T", false], [0, "_CPPv4eqRK5arrayRK5array", false]], "operator> (c++ function)": [[0, "_CPPv4I0Egt5array1TRK5array", false], [0, "_CPPv4I0Egt5arrayRK5array1T", false], [0, "_CPPv4gtRK5arrayRK5array", false]], "operator>= (c++ function)": [[0, "_CPPv4I0Ege5array1TRK5array", false], [0, "_CPPv4I0Ege5arrayRK5array1T", false], [0, "_CPPv4geRK5arrayRK5array", false]], "operator>> (c++ function)": [[0, "_CPPv4rsRK5arrayRK5array", false]], "operator^ (c++ function)": [[0, "_CPPv4eoRK5arrayRK5array", false]], "operator| (c++ function)": [[0, "_CPPv4orRK5arrayRK5array", false]], "operator|| (c++ function)": [[0, "_CPPv4ooRK5arrayRK5array", false]], "optimizer (class in mlx.optimizers)": [[474, "mlx.optimizers.Optimizer", false]], "outer (c++ function)": [[0, "_CPPv45outerRK5arrayRK5array14StreamOrDevice", false]], "outer() (in module mlx.core)": [[231, "mlx.core.outer", false]], "pad (c++ function)": [[0, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", false], [0, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", false], [0, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", false], [0, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", false]], "pad() (in module mlx.core)": [[232, "mlx.core.pad", false]], "parameters() (module method)": [[368, "mlx.nn.Module.parameters", false]], "partition (c++ function)": [[0, "_CPPv49partitionRK5arrayi14StreamOrDevice", false], [0, "_CPPv49partitionRK5arrayii14StreamOrDevice", false]], "partition() (in module mlx.core)": [[233, "mlx.core.partition", false]], "permutation() (in module mlx.core.random)": [[247, "mlx.core.random.permutation", false]], "power (c++ function)": [[0, "_CPPv45powerRK5arrayRK5array14StreamOrDevice", false]], "power() (in module mlx.core)": [[234, "mlx.core.power", false]], "prelu (class in mlx.nn)": [[379, "mlx.nn.PReLU", false], [436, "mlx.nn.prelu", false]], "prod (c++ function)": [[0, "_CPPv44prodRK5array14StreamOrDevice", false], [0, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv44prodRK5arrayb14StreamOrDevice", false], [0, "_CPPv44prodRK5arrayib14StreamOrDevice", false]], "prod() (array method)": [[63, "mlx.core.array.prod", false]], "prod() (in module mlx.core)": [[235, "mlx.core.prod", false]], "put_along_axis (c++ function)": [[0, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false]], "put_along_axis() (in module mlx.core)": [[236, "mlx.core.put_along_axis", false]], "qr() (in module mlx.core.linalg)": [[190, "mlx.core.linalg.qr", false]], "quantize (c++ function)": [[0, "_CPPv48quantizeRK5arrayii14StreamOrDevice", false]], "quantize() (in module mlx.core)": [[237, "mlx.core.quantize", false]], "quantize() (in module mlx.nn)": [[307, "mlx.nn.quantize", false]], "quantized_matmul (c++ function)": [[0, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", false]], "quantized_matmul() (in module mlx.core)": [[238, "mlx.core.quantized_matmul", false]], "quantizedembedding (class in mlx.nn)": [[380, "mlx.nn.QuantizedEmbedding", false]], "quantizedlinear (class in mlx.nn)": [[381, "mlx.nn.QuantizedLinear", false]], "radians (c++ function)": [[0, "_CPPv47radiansRK5array14StreamOrDevice", false]], "radians() (in module mlx.core)": [[239, "mlx.core.radians", false]], "randint() (in module mlx.core.random)": [[248, "mlx.core.random.randint", false]], "real (c++ function)": [[0, "_CPPv44realRK5array14StreamOrDevice", false]], "real() (in module mlx.core)": [[253, "mlx.core.real", false]], "reciprocal (c++ function)": [[0, "_CPPv410reciprocalRK5array14StreamOrDevice", false]], "reciprocal() (array method)": [[64, "mlx.core.array.reciprocal", false]], "reciprocal() (in module mlx.core)": [[254, "mlx.core.reciprocal", false]], "recv() (in module mlx.core.distributed)": [[125, "mlx.core.distributed.recv", false]], "recv_like() (in module mlx.core.distributed)": [[126, "mlx.core.distributed.recv_like", false]], "relu (class in mlx.nn)": [[384, "mlx.nn.ReLU", false], [437, "mlx.nn.relu", false]], "relu6 (class in mlx.nn)": [[385, "mlx.nn.ReLU6", false], [438, "mlx.nn.relu6", false]], "remainder (c++ function)": [[0, "_CPPv49remainderRK5arrayRK5array14StreamOrDevice", false]], "remainder() (in module mlx.core)": [[255, "mlx.core.remainder", false]], "repeat (c++ function)": [[0, "_CPPv46repeatRK5arrayi14StreamOrDevice", false], [0, "_CPPv46repeatRK5arrayii14StreamOrDevice", false]], "repeat() (in module mlx.core)": [[256, "mlx.core.repeat", false]], "reset_peak_memory() (in module mlx.core.metal)": [[215, "mlx.core.metal.reset_peak_memory", false]], "reshape (c++ function)": [[0, "_CPPv47reshapeRK5arrayNSt6vectorIiEE14StreamOrDevice", false]], "reshape() (array method)": [[65, "mlx.core.array.reshape", false]], "reshape() (in module mlx.core)": [[257, "mlx.core.reshape", false]], "rfft() (in module mlx.core.fft)": [[156, "mlx.core.fft.rfft", false]], "rfft2() (in module mlx.core.fft)": [[157, "mlx.core.fft.rfft2", false]], "rfftn() (in module mlx.core.fft)": [[158, "mlx.core.fft.rfftn", false]], "right_shift (c++ function)": [[0, "_CPPv411right_shiftRK5arrayRK5array14StreamOrDevice", false]], "right_shift() (in module mlx.core)": [[258, "mlx.core.right_shift", false]], "rms_norm() (in module mlx.core.fast)": [[144, "mlx.core.fast.rms_norm", false]], "rmsnorm (class in mlx.nn)": [[382, "mlx.nn.RMSNorm", false]], "rmsprop (class in mlx.optimizers)": [[466, "mlx.optimizers.RMSprop", false]], "rnn (class in mlx.nn)": [[383, "mlx.nn.RNN", false]], "roll (c++ function)": [[0, "_CPPv44rollRK5arrayRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", false], [0, "_CPPv44rollRK5arrayi14StreamOrDevice", false], [0, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv44rollRK5arrayii14StreamOrDevice", false]], "roll() (in module mlx.core)": [[259, "mlx.core.roll", false]], "rope (class in mlx.nn)": [[386, "mlx.nn.RoPE", false]], "rope() (in module mlx.core.fast)": [[145, "mlx.core.fast.rope", false]], "round (c++ function)": [[0, "_CPPv45roundRK5array14StreamOrDevice", false], [0, "_CPPv45roundRK5arrayi14StreamOrDevice", false]], "round() (array method)": [[66, "mlx.core.array.round", false]], "round() (in module mlx.core)": [[260, "mlx.core.round", false]], "rsqrt (c++ function)": [[0, "_CPPv45rsqrtRK5array14StreamOrDevice", false]], "rsqrt() (array method)": [[67, "mlx.core.array.rsqrt", false]], "rsqrt() (in module mlx.core)": [[261, "mlx.core.rsqrt", false]], "save() (in module mlx.core)": [[262, "mlx.core.save", false]], "save_gguf() (in module mlx.core)": [[263, "mlx.core.save_gguf", false]], "save_safetensors() (in module mlx.core)": [[264, "mlx.core.save_safetensors", false]], "save_weights() (module method)": [[369, "mlx.nn.Module.save_weights", false]], "savez() (in module mlx.core)": [[265, "mlx.core.savez", false]], "savez_compressed() (in module mlx.core)": [[266, "mlx.core.savez_compressed", false]], "scaled_dot_product_attention() (in module mlx.core.fast)": [[146, "mlx.core.fast.scaled_dot_product_attention", false]], "scatter (c++ function)": [[0, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "scatter_add (c++ function)": [[0, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "scatter_max (c++ function)": [[0, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "scatter_min (c++ function)": [[0, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "scatter_prod (c++ function)": [[0, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "seed() (in module mlx.core.random)": [[249, "mlx.core.random.seed", false]], "selu (class in mlx.nn)": [[387, "mlx.nn.SELU", false], [439, "mlx.nn.selu", false]], "send() (in module mlx.core.distributed)": [[127, "mlx.core.distributed.send", false]], "sequential (class in mlx.nn)": [[388, "mlx.nn.Sequential", false]], "set_cache_limit() (in module mlx.core.metal)": [[216, "mlx.core.metal.set_cache_limit", false]], "set_default_device() (in module mlx.core)": [[267, "mlx.core.set_default_device", false]], "set_default_stream() (in module mlx.core)": [[268, "mlx.core.set_default_stream", false]], "set_dtype() (module method)": [[370, "mlx.nn.Module.set_dtype", false]], "set_memory_limit() (in module mlx.core.metal)": [[217, "mlx.core.metal.set_memory_limit", false]], "set_wired_limit() (in module mlx.core.metal)": [[218, "mlx.core.metal.set_wired_limit", false]], "sgd (class in mlx.optimizers)": [[467, "mlx.optimizers.SGD", false]], "shape (array property)": [[68, "mlx.core.array.shape", false]], "sigmoid (c++ function)": [[0, "_CPPv47sigmoidRK5array14StreamOrDevice", false]], "sigmoid (class in mlx.nn)": [[390, "mlx.nn.Sigmoid", false], [440, "mlx.nn.sigmoid", false]], "sigmoid() (in module mlx.core)": [[269, "mlx.core.sigmoid", false]], "sign (c++ function)": [[0, "_CPPv44signRK5array14StreamOrDevice", false]], "sign() (in module mlx.core)": [[270, "mlx.core.sign", false]], "silu (class in mlx.nn)": [[389, "mlx.nn.SiLU", false], [441, "mlx.nn.silu", false]], "sin (c++ function)": [[0, "_CPPv43sinRK5array14StreamOrDevice", false]], "sin() (array method)": [[69, "mlx.core.array.sin", false]], "sin() (in module mlx.core)": [[271, "mlx.core.sin", false]], "sinh (c++ function)": [[0, "_CPPv44sinhRK5array14StreamOrDevice", false]], "sinh() (in module mlx.core)": [[272, "mlx.core.sinh", false]], "sinusoidalpositionalencoding (class in mlx.nn)": [[391, "mlx.nn.SinusoidalPositionalEncoding", false]], "size (array property)": [[70, "mlx.core.array.size", false]], "slice (c++ function)": [[0, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", false]], "slice_update (c++ function)": [[0, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", false]], "smooth_l1_loss (class in mlx.nn.losses)": [[433, "mlx.nn.losses.smooth_l1_loss", false]], "softmax (c++ function)": [[0, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv47softmaxRK5arrayb14StreamOrDevice", false], [0, "_CPPv47softmaxRK5arrayib14StreamOrDevice", false]], "softmax (class in mlx.nn)": [[392, "mlx.nn.Softmax", false], [442, "mlx.nn.softmax", false]], "softmax() (in module mlx.core)": [[273, "mlx.core.softmax", false]], "softmin (class in mlx.nn)": [[393, "mlx.nn.Softmin", false], [443, "mlx.nn.softmin", false]], "softplus (class in mlx.nn)": [[394, "mlx.nn.Softplus", false], [444, "mlx.nn.softplus", false]], "softshrink (class in mlx.nn)": [[395, "mlx.nn.Softshrink", false], [445, "mlx.nn.softshrink", false]], "softsign (class in mlx.nn)": [[396, "mlx.nn.Softsign", false]], "sort (c++ function)": [[0, "_CPPv44sortRK5array14StreamOrDevice", false], [0, "_CPPv44sortRK5arrayi14StreamOrDevice", false]], "sort() (in module mlx.core)": [[274, "mlx.core.sort", false]], "split (c++ function)": [[0, "_CPPv45splitRK5arrayRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", false], [0, "_CPPv45splitRK5arrayi14StreamOrDevice", false], [0, "_CPPv45splitRK5arrayii14StreamOrDevice", false]], "split() (array method)": [[71, "mlx.core.array.split", false]], "split() (in module mlx.core)": [[275, "mlx.core.split", false]], "split() (in module mlx.core.random)": [[250, "mlx.core.random.split", false]], "sqrt (c++ function)": [[0, "_CPPv44sqrtRK5array14StreamOrDevice", false]], "sqrt() (array method)": [[72, "mlx.core.array.sqrt", false]], "sqrt() (in module mlx.core)": [[276, "mlx.core.sqrt", false]], "square (c++ function)": [[0, "_CPPv46squareRK5array14StreamOrDevice", false]], "square() (array method)": [[73, "mlx.core.array.square", false]], "square() (in module mlx.core)": [[277, "mlx.core.square", false]], "squeeze (c++ function)": [[0, "_CPPv47squeezeRK5array14StreamOrDevice", false], [0, "_CPPv47squeezeRK5arrayRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv47squeezeRK5arrayi14StreamOrDevice", false]], "squeeze() (array method)": [[74, "mlx.core.array.squeeze", false]], "squeeze() (in module mlx.core)": [[278, "mlx.core.squeeze", false]], "stack (c++ function)": [[0, "_CPPv45stackRKNSt6vectorI5arrayEE14StreamOrDevice", false], [0, "_CPPv45stackRKNSt6vectorI5arrayEEi14StreamOrDevice", false]], "stack() (in module mlx.core)": [[279, "mlx.core.stack", false]], "start_capture() (in module mlx.core.metal)": [[219, "mlx.core.metal.start_capture", false]], "state (module property)": [[371, "mlx.nn.Module.state", false]], "state (optimizer property)": [[464, "mlx.optimizers.Optimizer.state", false]], "std (c++ function)": [[0, "_CPPv4StRK5array14StreamOrDevice", false], [0, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", false], [0, "_CPPv4StRK5arraybi14StreamOrDevice", false], [0, "_CPPv4StRK5arrayibi14StreamOrDevice", false]], "std() (array method)": [[75, "mlx.core.array.std", false]], "std() (in module mlx.core)": [[280, "mlx.core.std", false]], "step (class in mlx.nn)": [[397, "mlx.nn.Step", false], [446, "mlx.nn.step", false]], "step_decay() (in module mlx.optimizers)": [[472, "mlx.optimizers.step_decay", false]], "stop_capture() (in module mlx.core.metal)": [[220, "mlx.core.metal.stop_capture", false]], "stop_gradient (c++ function)": [[0, "_CPPv413stop_gradientRK5array14StreamOrDevice", false]], "stop_gradient() (in module mlx.core)": [[281, "mlx.core.stop_gradient", false]], "stream (class in mlx.core)": [[315, "mlx.core.Stream", false]], "stream() (in module mlx.core)": [[282, "mlx.core.stream", false]], "subtract (c++ function)": [[0, "_CPPv48subtractRK5arrayRK5array14StreamOrDevice", false]], "subtract() (in module mlx.core)": [[283, "mlx.core.subtract", false]], "sum (c++ function)": [[0, "_CPPv43sumRK5array14StreamOrDevice", false], [0, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv43sumRK5arrayb14StreamOrDevice", false], [0, "_CPPv43sumRK5arrayib14StreamOrDevice", false]], "sum() (array method)": [[76, "mlx.core.array.sum", false]], "sum() (in module mlx.core)": [[284, "mlx.core.sum", false]], "svd() (in module mlx.core.linalg)": [[191, "mlx.core.linalg.svd", false]], "swapaxes (c++ function)": [[0, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", false]], "swapaxes() (array method)": [[77, "mlx.core.array.swapaxes", false]], "swapaxes() (in module mlx.core)": [[285, "mlx.core.swapaxes", false]], "synchronize() (in module mlx.core)": [[286, "mlx.core.synchronize", false]], "t (array property)": [[31, "mlx.core.array.T", false]], "take (c++ function)": [[0, "_CPPv44takeRK5arrayRK5array14StreamOrDevice", false], [0, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv44takeRK5arrayi14StreamOrDevice", false], [0, "_CPPv44takeRK5arrayii14StreamOrDevice", false]], "take() (in module mlx.core)": [[287, "mlx.core.take", false]], "take_along_axis (c++ function)": [[0, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", false]], "take_along_axis() (in module mlx.core)": [[288, "mlx.core.take_along_axis", false]], "tan (c++ function)": [[0, "_CPPv43tanRK5array14StreamOrDevice", false]], "tan() (in module mlx.core)": [[289, "mlx.core.tan", false]], "tanh (c++ function)": [[0, "_CPPv44tanhRK5array14StreamOrDevice", false]], "tanh (class in mlx.nn)": [[398, "mlx.nn.Tanh", false], [447, "mlx.nn.tanh", false]], "tanh() (in module mlx.core)": [[290, "mlx.core.tanh", false]], "tensordot (c++ function)": [[0, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", false], [0, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", false]], "tensordot() (in module mlx.core)": [[291, "mlx.core.tensordot", false]], "tile (c++ function)": [[0, "_CPPv44tileRK5arrayNSt6vectorIiEE14StreamOrDevice", false]], "tile() (in module mlx.core)": [[292, "mlx.core.tile", false]], "tolist() (array method)": [[78, "mlx.core.array.tolist", false]], "topk (c++ function)": [[0, "_CPPv44topkRK5arrayi14StreamOrDevice", false], [0, "_CPPv44topkRK5arrayii14StreamOrDevice", false]], "topk() (in module mlx.core)": [[293, "mlx.core.topk", false]], "trace (c++ function)": [[0, "_CPPv45traceRK5array14StreamOrDevice", false], [0, "_CPPv45traceRK5arrayiii14StreamOrDevice", false], [0, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", false]], "trace() (in module mlx.core)": [[294, "mlx.core.trace", false]], "train() (module method)": [[372, "mlx.nn.Module.train", false]], "trainable_parameters() (module method)": [[373, "mlx.nn.Module.trainable_parameters", false]], "training (module property)": [[374, "mlx.nn.Module.training", false]], "transformer (class in mlx.nn)": [[399, "mlx.nn.Transformer", false]], "transpose (c++ function)": [[0, "_CPPv49transposeRK5array14StreamOrDevice", false], [0, "_CPPv49transposeRK5arrayNSt16initializer_listIiEE14StreamOrDevice", false], [0, "_CPPv49transposeRK5arrayNSt6vectorIiEE14StreamOrDevice", false]], "transpose() (array method)": [[79, "mlx.core.array.transpose", false]], "transpose() (in module mlx.core)": [[295, "mlx.core.transpose", false]], "tree_flatten() (in module mlx.utils)": [[310, "mlx.utils.tree_flatten", false]], "tree_map() (in module mlx.utils)": [[311, "mlx.utils.tree_map", false]], "tree_map_with_path() (in module mlx.utils)": [[312, "mlx.utils.tree_map_with_path", false]], "tree_reduce() (in module mlx.utils)": [[313, "mlx.utils.tree_reduce", false]], "tree_unflatten() (in module mlx.utils)": [[314, "mlx.utils.tree_unflatten", false]], "tri (c++ function)": [[0, "_CPPv43trii5Dtype14StreamOrDevice", false], [0, "_CPPv43triiii5Dtype14StreamOrDevice", false]], "tri() (in module mlx.core)": [[296, "mlx.core.tri", false]], "tri_inv() (in module mlx.core.linalg)": [[192, "mlx.core.linalg.tri_inv", false]], "tril (c++ function)": [[0, "_CPPv44tril5arrayi14StreamOrDevice", false]], "tril() (in module mlx.core)": [[297, "mlx.core.tril", false]], "triplet_loss (class in mlx.nn.losses)": [[434, "mlx.nn.losses.triplet_loss", false]], "triu (c++ function)": [[0, "_CPPv44triu5arrayi14StreamOrDevice", false]], "triu() (in module mlx.core)": [[298, "mlx.core.triu", false]], "truncated_normal() (in module mlx.core.random)": [[251, "mlx.core.random.truncated_normal", false]], "unfreeze() (module method)": [[375, "mlx.nn.Module.unfreeze", false]], "uniform() (in module mlx.core.random)": [[252, "mlx.core.random.uniform", false]], "uniform() (in module mlx.nn.init)": [[408, "mlx.nn.init.uniform", false]], "update() (module method)": [[376, "mlx.nn.Module.update", false]], "update() (optimizer method)": [[465, "mlx.optimizers.Optimizer.update", false]], "update_modules() (module method)": [[377, "mlx.nn.Module.update_modules", false]], "upsample (class in mlx.nn)": [[400, "mlx.nn.Upsample", false]], "value_and_grad() (in module mlx.core)": [[299, "mlx.core.value_and_grad", false]], "value_and_grad() (in module mlx.nn)": [[308, "mlx.nn.value_and_grad", false]], "var (c++ function)": [[0, "_CPPv43varRK5array14StreamOrDevice", false], [0, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", false], [0, "_CPPv43varRK5arraybi14StreamOrDevice", false], [0, "_CPPv43varRK5arrayibi14StreamOrDevice", false]], "var() (array method)": [[80, "mlx.core.array.var", false]], "var() (in module mlx.core)": [[300, "mlx.core.var", false]], "view (c++ function)": [[0, "_CPPv44viewRK5arrayRK5Dtype14StreamOrDevice", false]], "view() (array method)": [[81, "mlx.core.array.view", false]], "view() (in module mlx.core)": [[301, "mlx.core.view", false]], "vjp() (in module mlx.core)": [[302, "mlx.core.vjp", false]], "vmap() (in module mlx.core)": [[303, "mlx.core.vmap", false]], "where (c++ function)": [[0, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", false]], "where() (in module mlx.core)": [[304, "mlx.core.where", false]], "zeros (c++ function)": [[0, "_CPPv45zerosRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv45zerosRKNSt6vectorIiEE5Dtype14StreamOrDevice", false]], "zeros() (in module mlx.core)": [[305, "mlx.core.zeros", false]], "zeros_like (c++ function)": [[0, "_CPPv410zeros_likeRK5array14StreamOrDevice", false]], "zeros_like() (in module mlx.core)": [[306, "mlx.core.zeros_like", false]]}, "objects": {"": [[0, 0, 1, "_CPPv43absRK5array14StreamOrDevice", "abs"], [0, 1, 1, "_CPPv43absRK5array14StreamOrDevice", "abs::a"], [0, 1, 1, "_CPPv43absRK5array14StreamOrDevice", "abs::s"], [0, 0, 1, "_CPPv43addRK5arrayRK5array14StreamOrDevice", "add"], [0, 1, 1, "_CPPv43addRK5arrayRK5array14StreamOrDevice", "add::a"], [0, 1, 1, "_CPPv43addRK5arrayRK5array14StreamOrDevice", "add::b"], [0, 1, 1, "_CPPv43addRK5arrayRK5array14StreamOrDevice", "add::s"], [0, 0, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::a"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::alpha"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::b"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::beta"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::c"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::s"], [0, 0, 1, "_CPPv43allRK5array14StreamOrDevice", "all"], [0, 0, 1, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "all"], [0, 0, 1, "_CPPv43allRK5arrayb14StreamOrDevice", "all"], [0, 0, 1, "_CPPv43allRK5arrayib14StreamOrDevice", "all"], [0, 1, 1, "_CPPv43allRK5array14StreamOrDevice", "all::a"], [0, 1, 1, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "all::a"], [0, 1, 1, "_CPPv43allRK5arrayb14StreamOrDevice", "all::a"], [0, 1, 1, "_CPPv43allRK5arrayib14StreamOrDevice", "all::a"], [0, 1, 1, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "all::axes"], [0, 1, 1, "_CPPv43allRK5arrayib14StreamOrDevice", "all::axis"], [0, 1, 1, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "all::keepdims"], [0, 1, 1, "_CPPv43allRK5arrayb14StreamOrDevice", "all::keepdims"], [0, 1, 1, "_CPPv43allRK5arrayib14StreamOrDevice", "all::keepdims"], [0, 1, 1, "_CPPv43allRK5array14StreamOrDevice", "all::s"], [0, 1, 1, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "all::s"], [0, 1, 1, "_CPPv43allRK5arrayb14StreamOrDevice", "all::s"], [0, 1, 1, "_CPPv43allRK5arrayib14StreamOrDevice", "all::s"], [0, 0, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::a"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::atol"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::b"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::equal_nan"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::rtol"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::s"], [0, 0, 1, "_CPPv43anyRK5array14StreamOrDevice", "any"], [0, 0, 1, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "any"], [0, 0, 1, "_CPPv43anyRK5arrayb14StreamOrDevice", "any"], [0, 0, 1, "_CPPv43anyRK5arrayib14StreamOrDevice", "any"], [0, 1, 1, "_CPPv43anyRK5array14StreamOrDevice", "any::a"], [0, 1, 1, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "any::a"], [0, 1, 1, "_CPPv43anyRK5arrayb14StreamOrDevice", "any::a"], [0, 1, 1, "_CPPv43anyRK5arrayib14StreamOrDevice", "any::a"], [0, 1, 1, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "any::axes"], [0, 1, 1, "_CPPv43anyRK5arrayib14StreamOrDevice", "any::axis"], [0, 1, 1, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "any::keepdims"], [0, 1, 1, "_CPPv43anyRK5arrayb14StreamOrDevice", "any::keepdims"], [0, 1, 1, "_CPPv43anyRK5arrayib14StreamOrDevice", "any::keepdims"], [0, 1, 1, "_CPPv43anyRK5array14StreamOrDevice", "any::s"], [0, 1, 1, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "any::s"], [0, 1, 1, "_CPPv43anyRK5arrayb14StreamOrDevice", "any::s"], [0, 1, 1, "_CPPv43anyRK5arrayib14StreamOrDevice", "any::s"], [0, 0, 1, "_CPPv46aranged14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46aranged5Dtype14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangedd14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangedd5Dtype14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangeddd14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangei14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangeii14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangeiii14StreamOrDevice", "arange"], [0, 1, 1, "_CPPv46aranged5Dtype14StreamOrDevice", "arange::dtype"], [0, 1, 1, "_CPPv46arangedd5Dtype14StreamOrDevice", "arange::dtype"], [0, 1, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange::dtype"], [0, 1, 1, "_CPPv46aranged14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46aranged5Dtype14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangedd14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangedd5Dtype14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangeddd14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangei14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangeii14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangeiii14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangedd14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangedd5Dtype14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangeddd14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangeii14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangeiii14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangeddd14StreamOrDevice", "arange::step"], [0, 1, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange::step"], [0, 1, 1, "_CPPv46arangeiii14StreamOrDevice", "arange::step"], [0, 1, 1, "_CPPv46aranged14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46aranged5Dtype14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangedd14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangedd5Dtype14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangeddd14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangei14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangeii14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangeiii14StreamOrDevice", "arange::stop"], [0, 0, 1, "_CPPv46arccosRK5array14StreamOrDevice", "arccos"], [0, 1, 1, "_CPPv46arccosRK5array14StreamOrDevice", "arccos::a"], [0, 1, 1, "_CPPv46arccosRK5array14StreamOrDevice", "arccos::s"], [0, 0, 1, "_CPPv47arccoshRK5array14StreamOrDevice", "arccosh"], [0, 1, 1, "_CPPv47arccoshRK5array14StreamOrDevice", "arccosh::a"], [0, 1, 1, "_CPPv47arccoshRK5array14StreamOrDevice", "arccosh::s"], [0, 0, 1, "_CPPv46arcsinRK5array14StreamOrDevice", "arcsin"], [0, 1, 1, "_CPPv46arcsinRK5array14StreamOrDevice", "arcsin::a"], [0, 1, 1, "_CPPv46arcsinRK5array14StreamOrDevice", "arcsin::s"], [0, 0, 1, "_CPPv47arcsinhRK5array14StreamOrDevice", "arcsinh"], [0, 1, 1, "_CPPv47arcsinhRK5array14StreamOrDevice", "arcsinh::a"], [0, 1, 1, "_CPPv47arcsinhRK5array14StreamOrDevice", "arcsinh::s"], [0, 0, 1, "_CPPv46arctanRK5array14StreamOrDevice", "arctan"], [0, 0, 1, "_CPPv47arctan2RK5arrayRK5array14StreamOrDevice", "arctan2"], [0, 1, 1, "_CPPv47arctan2RK5arrayRK5array14StreamOrDevice", "arctan2::a"], [0, 1, 1, "_CPPv47arctan2RK5arrayRK5array14StreamOrDevice", "arctan2::b"], [0, 1, 1, "_CPPv47arctan2RK5arrayRK5array14StreamOrDevice", "arctan2::s"], [0, 1, 1, "_CPPv46arctanRK5array14StreamOrDevice", "arctan::a"], [0, 1, 1, "_CPPv46arctanRK5array14StreamOrDevice", "arctan::s"], [0, 0, 1, "_CPPv47arctanhRK5array14StreamOrDevice", "arctanh"], [0, 1, 1, "_CPPv47arctanhRK5array14StreamOrDevice", "arctanh::a"], [0, 1, 1, "_CPPv47arctanhRK5array14StreamOrDevice", "arctanh::s"], [0, 0, 1, "_CPPv46argmaxRK5array14StreamOrDevice", "argmax"], [0, 0, 1, "_CPPv46argmaxRK5arrayb14StreamOrDevice", "argmax"], [0, 0, 1, "_CPPv46argmaxRK5arrayib14StreamOrDevice", "argmax"], [0, 1, 1, "_CPPv46argmaxRK5array14StreamOrDevice", "argmax::a"], [0, 1, 1, "_CPPv46argmaxRK5arrayb14StreamOrDevice", "argmax::a"], [0, 1, 1, "_CPPv46argmaxRK5arrayib14StreamOrDevice", "argmax::a"], [0, 1, 1, "_CPPv46argmaxRK5arrayib14StreamOrDevice", "argmax::axis"], [0, 1, 1, "_CPPv46argmaxRK5arrayb14StreamOrDevice", "argmax::keepdims"], [0, 1, 1, "_CPPv46argmaxRK5arrayib14StreamOrDevice", "argmax::keepdims"], [0, 1, 1, "_CPPv46argmaxRK5array14StreamOrDevice", "argmax::s"], [0, 1, 1, "_CPPv46argmaxRK5arrayb14StreamOrDevice", "argmax::s"], [0, 1, 1, "_CPPv46argmaxRK5arrayib14StreamOrDevice", "argmax::s"], [0, 0, 1, "_CPPv46argminRK5array14StreamOrDevice", "argmin"], [0, 0, 1, "_CPPv46argminRK5arrayb14StreamOrDevice", "argmin"], [0, 0, 1, "_CPPv46argminRK5arrayib14StreamOrDevice", "argmin"], [0, 1, 1, "_CPPv46argminRK5array14StreamOrDevice", "argmin::a"], [0, 1, 1, "_CPPv46argminRK5arrayb14StreamOrDevice", "argmin::a"], [0, 1, 1, "_CPPv46argminRK5arrayib14StreamOrDevice", "argmin::a"], [0, 1, 1, "_CPPv46argminRK5arrayib14StreamOrDevice", "argmin::axis"], [0, 1, 1, "_CPPv46argminRK5arrayb14StreamOrDevice", "argmin::keepdims"], [0, 1, 1, "_CPPv46argminRK5arrayib14StreamOrDevice", "argmin::keepdims"], [0, 1, 1, "_CPPv46argminRK5array14StreamOrDevice", "argmin::s"], [0, 1, 1, "_CPPv46argminRK5arrayb14StreamOrDevice", "argmin::s"], [0, 1, 1, "_CPPv46argminRK5arrayib14StreamOrDevice", "argmin::s"], [0, 0, 1, "_CPPv412argpartitionRK5arrayi14StreamOrDevice", "argpartition"], [0, 0, 1, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", "argpartition"], [0, 1, 1, "_CPPv412argpartitionRK5arrayi14StreamOrDevice", "argpartition::a"], [0, 1, 1, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", "argpartition::a"], [0, 1, 1, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", "argpartition::axis"], [0, 1, 1, "_CPPv412argpartitionRK5arrayi14StreamOrDevice", "argpartition::kth"], [0, 1, 1, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", "argpartition::kth"], [0, 1, 1, "_CPPv412argpartitionRK5arrayi14StreamOrDevice", "argpartition::s"], [0, 1, 1, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", "argpartition::s"], [0, 0, 1, "_CPPv47argsortRK5array14StreamOrDevice", "argsort"], [0, 0, 1, "_CPPv47argsortRK5arrayi14StreamOrDevice", "argsort"], [0, 1, 1, "_CPPv47argsortRK5array14StreamOrDevice", "argsort::a"], [0, 1, 1, "_CPPv47argsortRK5arrayi14StreamOrDevice", "argsort::a"], [0, 1, 1, "_CPPv47argsortRK5arrayi14StreamOrDevice", "argsort::axis"], [0, 1, 1, "_CPPv47argsortRK5array14StreamOrDevice", "argsort::s"], [0, 1, 1, "_CPPv47argsortRK5arrayi14StreamOrDevice", "argsort::s"], [0, 0, 1, "_CPPv411array_equalRK5arrayRK5array14StreamOrDevice", "array_equal"], [0, 0, 1, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", "array_equal"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5array14StreamOrDevice", "array_equal::a"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", "array_equal::a"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5array14StreamOrDevice", "array_equal::b"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", "array_equal::b"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", "array_equal::equal_nan"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5array14StreamOrDevice", "array_equal::s"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", "array_equal::s"], [0, 0, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided"], [0, 1, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided::a"], [0, 1, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided::offset"], [0, 1, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided::s"], [0, 1, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided::shape"], [0, 1, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided::strides"], [0, 0, 1, "_CPPv46astype5array5Dtype14StreamOrDevice", "astype"], [0, 1, 1, "_CPPv46astype5array5Dtype14StreamOrDevice", "astype::a"], [0, 1, 1, "_CPPv46astype5array5Dtype14StreamOrDevice", "astype::dtype"], [0, 1, 1, "_CPPv46astype5array5Dtype14StreamOrDevice", "astype::s"], [0, 0, 1, "_CPPv410atleast_1dRK5array14StreamOrDevice", "atleast_1d"], [0, 0, 1, "_CPPv410atleast_1dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_1d"], [0, 1, 1, "_CPPv410atleast_1dRK5array14StreamOrDevice", "atleast_1d::a"], [0, 1, 1, "_CPPv410atleast_1dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_1d::a"], [0, 1, 1, "_CPPv410atleast_1dRK5array14StreamOrDevice", "atleast_1d::s"], [0, 1, 1, "_CPPv410atleast_1dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_1d::s"], [0, 0, 1, "_CPPv410atleast_2dRK5array14StreamOrDevice", "atleast_2d"], [0, 0, 1, "_CPPv410atleast_2dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_2d"], [0, 1, 1, "_CPPv410atleast_2dRK5array14StreamOrDevice", "atleast_2d::a"], [0, 1, 1, "_CPPv410atleast_2dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_2d::a"], [0, 1, 1, "_CPPv410atleast_2dRK5array14StreamOrDevice", "atleast_2d::s"], [0, 1, 1, "_CPPv410atleast_2dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_2d::s"], [0, 0, 1, "_CPPv410atleast_3dRK5array14StreamOrDevice", "atleast_3d"], [0, 0, 1, "_CPPv410atleast_3dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_3d"], [0, 1, 1, "_CPPv410atleast_3dRK5array14StreamOrDevice", "atleast_3d::a"], [0, 1, 1, "_CPPv410atleast_3dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_3d::a"], [0, 1, 1, "_CPPv410atleast_3dRK5array14StreamOrDevice", "atleast_3d::s"], [0, 1, 1, "_CPPv410atleast_3dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_3d::s"], [0, 0, 1, "_CPPv411bitwise_andRK5arrayRK5array14StreamOrDevice", "bitwise_and"], [0, 1, 1, "_CPPv411bitwise_andRK5arrayRK5array14StreamOrDevice", "bitwise_and::a"], [0, 1, 1, "_CPPv411bitwise_andRK5arrayRK5array14StreamOrDevice", "bitwise_and::b"], [0, 1, 1, "_CPPv411bitwise_andRK5arrayRK5array14StreamOrDevice", "bitwise_and::s"], [0, 0, 1, "_CPPv410bitwise_orRK5arrayRK5array14StreamOrDevice", "bitwise_or"], [0, 1, 1, "_CPPv410bitwise_orRK5arrayRK5array14StreamOrDevice", "bitwise_or::a"], [0, 1, 1, "_CPPv410bitwise_orRK5arrayRK5array14StreamOrDevice", "bitwise_or::b"], [0, 1, 1, "_CPPv410bitwise_orRK5arrayRK5array14StreamOrDevice", "bitwise_or::s"], [0, 0, 1, "_CPPv411bitwise_xorRK5arrayRK5array14StreamOrDevice", "bitwise_xor"], [0, 1, 1, "_CPPv411bitwise_xorRK5arrayRK5array14StreamOrDevice", "bitwise_xor::a"], [0, 1, 1, "_CPPv411bitwise_xorRK5arrayRK5array14StreamOrDevice", "bitwise_xor::b"], [0, 1, 1, "_CPPv411bitwise_xorRK5arrayRK5array14StreamOrDevice", "bitwise_xor::s"], [0, 0, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::a"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::b"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::block_size"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::mask_lhs"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::mask_out"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::mask_rhs"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::s"], [0, 0, 1, "_CPPv416broadcast_arraysRKNSt6vectorI5arrayEE14StreamOrDevice", "broadcast_arrays"], [0, 1, 1, "_CPPv416broadcast_arraysRKNSt6vectorI5arrayEE14StreamOrDevice", "broadcast_arrays::inputs"], [0, 1, 1, "_CPPv416broadcast_arraysRKNSt6vectorI5arrayEE14StreamOrDevice", "broadcast_arrays::s"], [0, 0, 1, "_CPPv412broadcast_toRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "broadcast_to"], [0, 1, 1, "_CPPv412broadcast_toRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "broadcast_to::a"], [0, 1, 1, "_CPPv412broadcast_toRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "broadcast_to::s"], [0, 1, 1, "_CPPv412broadcast_toRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "broadcast_to::shape"], [0, 0, 1, "_CPPv44ceilRK5array14StreamOrDevice", "ceil"], [0, 1, 1, "_CPPv44ceilRK5array14StreamOrDevice", "ceil::a"], [0, 1, 1, "_CPPv44ceilRK5array14StreamOrDevice", "ceil::s"], [0, 0, 1, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", "clip"], [0, 1, 1, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", "clip::a"], [0, 1, 1, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", "clip::a_max"], [0, 1, 1, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", "clip::a_min"], [0, 1, 1, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", "clip::s"], [0, 0, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEE14StreamOrDevice", "concatenate"], [0, 0, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEEi14StreamOrDevice", "concatenate"], [0, 1, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEE14StreamOrDevice", "concatenate::arrays"], [0, 1, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEEi14StreamOrDevice", "concatenate::arrays"], [0, 1, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEEi14StreamOrDevice", "concatenate::axis"], [0, 1, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEE14StreamOrDevice", "concatenate::s"], [0, 1, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEEi14StreamOrDevice", "concatenate::s"], [0, 0, 1, "_CPPv49conjugateRK5array14StreamOrDevice", "conjugate"], [0, 1, 1, "_CPPv49conjugateRK5array14StreamOrDevice", "conjugate::a"], [0, 1, 1, "_CPPv49conjugateRK5array14StreamOrDevice", "conjugate::s"], [0, 0, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::dilation"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::groups"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::input"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::padding"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::s"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::stride"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::weight"], [0, 0, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::dilation"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::groups"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::input"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::padding"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::s"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::stride"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::weight"], [0, 0, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::dilation"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::groups"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::input"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::padding"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::s"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::stride"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::weight"], [0, 0, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general"], [0, 0, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::flip"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::flip"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::groups"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::groups"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::input"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::input"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::input_dilation"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::input_dilation"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::kernel_dilation"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::kernel_dilation"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::padding"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::padding_hi"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::padding_lo"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::s"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::s"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::stride"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::stride"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::weight"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::weight"], [0, 0, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::dilation"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::groups"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::input"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::padding"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::s"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::stride"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::weight"], [0, 0, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::dilation"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::groups"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::input"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::padding"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::s"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::stride"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::weight"], [0, 0, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::dilation"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::groups"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::input"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::padding"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::s"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::stride"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::weight"], [0, 0, 1, "_CPPv44copy5array14StreamOrDevice", "copy"], [0, 1, 1, "_CPPv44copy5array14StreamOrDevice", "copy::a"], [0, 1, 1, "_CPPv44copy5array14StreamOrDevice", "copy::s"], [0, 0, 1, "_CPPv43cosRK5array14StreamOrDevice", "cos"], [0, 1, 1, "_CPPv43cosRK5array14StreamOrDevice", "cos::a"], [0, 1, 1, "_CPPv43cosRK5array14StreamOrDevice", "cos::s"], [0, 0, 1, "_CPPv44coshRK5array14StreamOrDevice", "cosh"], [0, 1, 1, "_CPPv44coshRK5array14StreamOrDevice", "cosh::a"], [0, 1, 1, "_CPPv44coshRK5array14StreamOrDevice", "cosh::s"], [0, 0, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax"], [0, 1, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax::a"], [0, 1, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax::axis"], [0, 1, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax::inclusive"], [0, 1, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax::reverse"], [0, 1, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax::s"], [0, 0, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin"], [0, 1, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin::a"], [0, 1, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin::axis"], [0, 1, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin::inclusive"], [0, 1, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin::reverse"], [0, 1, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin::s"], [0, 0, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod"], [0, 1, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod::a"], [0, 1, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod::axis"], [0, 1, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod::inclusive"], [0, 1, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod::reverse"], [0, 1, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod::s"], [0, 0, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum"], [0, 1, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum::a"], [0, 1, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum::axis"], [0, 1, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum::inclusive"], [0, 1, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum::reverse"], [0, 1, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum::s"], [0, 0, 1, "_CPPv47degreesRK5array14StreamOrDevice", "degrees"], [0, 1, 1, "_CPPv47degreesRK5array14StreamOrDevice", "degrees::a"], [0, 1, 1, "_CPPv47degreesRK5array14StreamOrDevice", "degrees::s"], [0, 0, 1, "_CPPv47dependsRKNSt6vectorI5arrayEERKNSt6vectorI5arrayEE", "depends"], [0, 1, 1, "_CPPv47dependsRKNSt6vectorI5arrayEERKNSt6vectorI5arrayEE", "depends::dependencies"], [0, 1, 1, "_CPPv47dependsRKNSt6vectorI5arrayEERKNSt6vectorI5arrayEE", "depends::inputs"], [0, 0, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::biases"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::bits"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::group_size"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::s"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::scales"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::w"], [0, 0, 1, "_CPPv44diagRK5arrayi14StreamOrDevice", "diag"], [0, 1, 1, "_CPPv44diagRK5arrayi14StreamOrDevice", "diag::a"], [0, 1, 1, "_CPPv44diagRK5arrayi14StreamOrDevice", "diag::k"], [0, 1, 1, "_CPPv44diagRK5arrayi14StreamOrDevice", "diag::s"], [0, 0, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal"], [0, 1, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal::a"], [0, 1, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal::axis1"], [0, 1, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal::axis2"], [0, 1, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal::offset"], [0, 1, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal::s"], [0, 0, 1, "_CPPv46divideRK5arrayRK5array14StreamOrDevice", "divide"], [0, 1, 1, "_CPPv46divideRK5arrayRK5array14StreamOrDevice", "divide::a"], [0, 1, 1, "_CPPv46divideRK5arrayRK5array14StreamOrDevice", "divide::b"], [0, 1, 1, "_CPPv46divideRK5arrayRK5array14StreamOrDevice", "divide::s"], [0, 0, 1, "_CPPv46divmodRK5arrayRK5array14StreamOrDevice", "divmod"], [0, 1, 1, "_CPPv46divmodRK5arrayRK5array14StreamOrDevice", "divmod::a"], [0, 1, 1, "_CPPv46divmodRK5arrayRK5array14StreamOrDevice", "divmod::b"], [0, 1, 1, "_CPPv46divmodRK5arrayRK5array14StreamOrDevice", "divmod::s"], [0, 0, 1, "_CPPv45equalRK5arrayRK5array14StreamOrDevice", "equal"], [0, 1, 1, "_CPPv45equalRK5arrayRK5array14StreamOrDevice", "equal::a"], [0, 1, 1, "_CPPv45equalRK5arrayRK5array14StreamOrDevice", "equal::b"], [0, 1, 1, "_CPPv45equalRK5arrayRK5array14StreamOrDevice", "equal::s"], [0, 0, 1, "_CPPv43erfRK5array14StreamOrDevice", "erf"], [0, 1, 1, "_CPPv43erfRK5array14StreamOrDevice", "erf::a"], [0, 1, 1, "_CPPv43erfRK5array14StreamOrDevice", "erf::s"], [0, 0, 1, "_CPPv46erfinvRK5array14StreamOrDevice", "erfinv"], [0, 1, 1, "_CPPv46erfinvRK5array14StreamOrDevice", "erfinv::a"], [0, 1, 1, "_CPPv46erfinvRK5array14StreamOrDevice", "erfinv::s"], [0, 0, 1, "_CPPv43expRK5array14StreamOrDevice", "exp"], [0, 1, 1, "_CPPv43expRK5array14StreamOrDevice", "exp::a"], [0, 1, 1, "_CPPv43expRK5array14StreamOrDevice", "exp::s"], [0, 0, 1, "_CPPv411expand_dimsRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "expand_dims"], [0, 0, 1, "_CPPv411expand_dimsRK5arrayi14StreamOrDevice", "expand_dims"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "expand_dims::a"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayi14StreamOrDevice", "expand_dims::a"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "expand_dims::axes"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayi14StreamOrDevice", "expand_dims::axis"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "expand_dims::s"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayi14StreamOrDevice", "expand_dims::s"], [0, 0, 1, "_CPPv45expm1RK5array14StreamOrDevice", "expm1"], [0, 1, 1, "_CPPv45expm1RK5array14StreamOrDevice", "expm1::a"], [0, 1, 1, "_CPPv45expm1RK5array14StreamOrDevice", "expm1::s"], [0, 0, 1, "_CPPv43eyei14StreamOrDevice", "eye"], [0, 0, 1, "_CPPv43eyei5Dtype14StreamOrDevice", "eye"], [0, 0, 1, "_CPPv43eyeii14StreamOrDevice", "eye"], [0, 0, 1, "_CPPv43eyeiii14StreamOrDevice", "eye"], [0, 0, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye"], [0, 1, 1, "_CPPv43eyei5Dtype14StreamOrDevice", "eye::dtype"], [0, 1, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye::dtype"], [0, 1, 1, "_CPPv43eyeiii14StreamOrDevice", "eye::k"], [0, 1, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye::k"], [0, 1, 1, "_CPPv43eyeii14StreamOrDevice", "eye::m"], [0, 1, 1, "_CPPv43eyeiii14StreamOrDevice", "eye::m"], [0, 1, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye::m"], [0, 1, 1, "_CPPv43eyei14StreamOrDevice", "eye::n"], [0, 1, 1, "_CPPv43eyei5Dtype14StreamOrDevice", "eye::n"], [0, 1, 1, "_CPPv43eyeii14StreamOrDevice", "eye::n"], [0, 1, 1, "_CPPv43eyeiii14StreamOrDevice", "eye::n"], [0, 1, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye::n"], [0, 1, 1, "_CPPv43eyei14StreamOrDevice", "eye::s"], [0, 1, 1, "_CPPv43eyei5Dtype14StreamOrDevice", "eye::s"], [0, 1, 1, "_CPPv43eyeii14StreamOrDevice", "eye::s"], [0, 1, 1, "_CPPv43eyeiii14StreamOrDevice", "eye::s"], [0, 1, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye::s"], [0, 0, 1, "_CPPv47flattenRK5array14StreamOrDevice", "flatten"], [0, 0, 1, "_CPPv47flattenRK5arrayii14StreamOrDevice", "flatten"], [0, 1, 1, "_CPPv47flattenRK5array14StreamOrDevice", "flatten::a"], [0, 1, 1, "_CPPv47flattenRK5arrayii14StreamOrDevice", "flatten::a"], [0, 1, 1, "_CPPv47flattenRK5arrayii14StreamOrDevice", "flatten::end_axis"], [0, 1, 1, "_CPPv47flattenRK5array14StreamOrDevice", "flatten::s"], [0, 1, 1, "_CPPv47flattenRK5arrayii14StreamOrDevice", "flatten::s"], [0, 1, 1, "_CPPv47flattenRK5arrayii14StreamOrDevice", "flatten::start_axis"], [0, 0, 1, "_CPPv45floorRK5array14StreamOrDevice", "floor"], [0, 1, 1, "_CPPv45floorRK5array14StreamOrDevice", "floor::a"], [0, 1, 1, "_CPPv45floorRK5array14StreamOrDevice", "floor::s"], [0, 0, 1, "_CPPv412floor_divideRK5arrayRK5array14StreamOrDevice", "floor_divide"], [0, 1, 1, "_CPPv412floor_divideRK5arrayRK5array14StreamOrDevice", "floor_divide::a"], [0, 1, 1, "_CPPv412floor_divideRK5arrayRK5array14StreamOrDevice", "floor_divide::b"], [0, 1, 1, "_CPPv412floor_divideRK5arrayRK5array14StreamOrDevice", "floor_divide::s"], [0, 0, 1, "_CPPv44fullNSt6vectorIiEE5array14StreamOrDevice", "full"], [0, 0, 1, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", "full"], [0, 0, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", "full"], [0, 0, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full"], [0, 2, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", "full::T"], [0, 2, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full::T"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", "full::dtype"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full::dtype"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array14StreamOrDevice", "full::s"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", "full::s"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", "full::s"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full::s"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array14StreamOrDevice", "full::shape"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", "full::shape"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", "full::shape"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full::shape"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", "full::val"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full::val"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array14StreamOrDevice", "full::vals"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", "full::vals"], [0, 0, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather"], [0, 0, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather"], [0, 1, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather::a"], [0, 1, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather::a"], [0, 1, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather::axes"], [0, 1, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather::axis"], [0, 1, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather::indices"], [0, 1, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather::indices"], [0, 1, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather::s"], [0, 1, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather::s"], [0, 1, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather::slice_sizes"], [0, 1, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather::slice_sizes"], [0, 0, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm"], [0, 1, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm::a"], [0, 1, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm::b"], [0, 1, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm::lhs_indices"], [0, 1, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm::rhs_indices"], [0, 1, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm::s"], [0, 0, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::biases"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::bits"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::group_size"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::lhs_indices"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::rhs_indices"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::s"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::scales"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::transpose"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::w"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::x"], [0, 0, 1, "_CPPv47greaterRK5arrayRK5array14StreamOrDevice", "greater"], [0, 1, 1, "_CPPv47greaterRK5arrayRK5array14StreamOrDevice", "greater::a"], [0, 1, 1, "_CPPv47greaterRK5arrayRK5array14StreamOrDevice", "greater::b"], [0, 1, 1, "_CPPv47greaterRK5arrayRK5array14StreamOrDevice", "greater::s"], [0, 0, 1, "_CPPv413greater_equalRK5arrayRK5array14StreamOrDevice", "greater_equal"], [0, 1, 1, "_CPPv413greater_equalRK5arrayRK5array14StreamOrDevice", "greater_equal::a"], [0, 1, 1, "_CPPv413greater_equalRK5arrayRK5array14StreamOrDevice", "greater_equal::b"], [0, 1, 1, "_CPPv413greater_equalRK5arrayRK5array14StreamOrDevice", "greater_equal::s"], [0, 0, 1, "_CPPv418hadamard_transformRK5arrayNSt8optionalIfEE14StreamOrDevice", "hadamard_transform"], [0, 1, 1, "_CPPv418hadamard_transformRK5arrayNSt8optionalIfEE14StreamOrDevice", "hadamard_transform::a"], [0, 1, 1, "_CPPv418hadamard_transformRK5arrayNSt8optionalIfEE14StreamOrDevice", "hadamard_transform::s"], [0, 1, 1, "_CPPv418hadamard_transformRK5arrayNSt8optionalIfEE14StreamOrDevice", "hadamard_transform::scale"], [0, 0, 1, "_CPPv48identityi14StreamOrDevice", "identity"], [0, 0, 1, "_CPPv48identityi5Dtype14StreamOrDevice", "identity"], [0, 1, 1, "_CPPv48identityi5Dtype14StreamOrDevice", "identity::dtype"], [0, 1, 1, "_CPPv48identityi14StreamOrDevice", "identity::n"], [0, 1, 1, "_CPPv48identityi5Dtype14StreamOrDevice", "identity::n"], [0, 1, 1, "_CPPv48identityi14StreamOrDevice", "identity::s"], [0, 1, 1, "_CPPv48identityi5Dtype14StreamOrDevice", "identity::s"], [0, 0, 1, "_CPPv44imagRK5array14StreamOrDevice", "imag"], [0, 1, 1, "_CPPv44imagRK5array14StreamOrDevice", "imag::a"], [0, 1, 1, "_CPPv44imagRK5array14StreamOrDevice", "imag::s"], [0, 0, 1, "_CPPv45innerRK5arrayRK5array14StreamOrDevice", "inner"], [0, 1, 1, "_CPPv45innerRK5arrayRK5array14StreamOrDevice", "inner::a"], [0, 1, 1, "_CPPv45innerRK5arrayRK5array14StreamOrDevice", "inner::b"], [0, 1, 1, "_CPPv45innerRK5arrayRK5array14StreamOrDevice", "inner::s"], [0, 0, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::a"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::atol"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::b"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::equal_nan"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::rtol"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::s"], [0, 0, 1, "_CPPv48isfiniteRK5array14StreamOrDevice", "isfinite"], [0, 1, 1, "_CPPv48isfiniteRK5array14StreamOrDevice", "isfinite::a"], [0, 1, 1, "_CPPv48isfiniteRK5array14StreamOrDevice", "isfinite::s"], [0, 0, 1, "_CPPv45isinfRK5array14StreamOrDevice", "isinf"], [0, 1, 1, "_CPPv45isinfRK5array14StreamOrDevice", "isinf::a"], [0, 1, 1, "_CPPv45isinfRK5array14StreamOrDevice", "isinf::s"], [0, 0, 1, "_CPPv45isnanRK5array14StreamOrDevice", "isnan"], [0, 1, 1, "_CPPv45isnanRK5array14StreamOrDevice", "isnan::a"], [0, 1, 1, "_CPPv45isnanRK5array14StreamOrDevice", "isnan::s"], [0, 0, 1, "_CPPv48isneginfRK5array14StreamOrDevice", "isneginf"], [0, 1, 1, "_CPPv48isneginfRK5array14StreamOrDevice", "isneginf::a"], [0, 1, 1, "_CPPv48isneginfRK5array14StreamOrDevice", "isneginf::s"], [0, 0, 1, "_CPPv48isposinfRK5array14StreamOrDevice", "isposinf"], [0, 1, 1, "_CPPv48isposinfRK5array14StreamOrDevice", "isposinf::a"], [0, 1, 1, "_CPPv48isposinfRK5array14StreamOrDevice", "isposinf::s"], [0, 0, 1, "_CPPv410left_shiftRK5arrayRK5array14StreamOrDevice", "left_shift"], [0, 1, 1, "_CPPv410left_shiftRK5arrayRK5array14StreamOrDevice", "left_shift::a"], [0, 1, 1, "_CPPv410left_shiftRK5arrayRK5array14StreamOrDevice", "left_shift::b"], [0, 1, 1, "_CPPv410left_shiftRK5arrayRK5array14StreamOrDevice", "left_shift::s"], [0, 0, 1, "_CPPv44lessRK5arrayRK5array14StreamOrDevice", "less"], [0, 1, 1, "_CPPv44lessRK5arrayRK5array14StreamOrDevice", "less::a"], [0, 1, 1, "_CPPv44lessRK5arrayRK5array14StreamOrDevice", "less::b"], [0, 1, 1, "_CPPv44lessRK5arrayRK5array14StreamOrDevice", "less::s"], [0, 0, 1, "_CPPv410less_equalRK5arrayRK5array14StreamOrDevice", "less_equal"], [0, 1, 1, "_CPPv410less_equalRK5arrayRK5array14StreamOrDevice", "less_equal::a"], [0, 1, 1, "_CPPv410less_equalRK5arrayRK5array14StreamOrDevice", "less_equal::b"], [0, 1, 1, "_CPPv410less_equalRK5arrayRK5array14StreamOrDevice", "less_equal::s"], [0, 0, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace"], [0, 1, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace::dtype"], [0, 1, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace::num"], [0, 1, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace::s"], [0, 1, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace::start"], [0, 1, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace::stop"], [0, 0, 1, "_CPPv43logRK5array14StreamOrDevice", "log"], [0, 0, 1, "_CPPv45log10RK5array14StreamOrDevice", "log10"], [0, 1, 1, "_CPPv45log10RK5array14StreamOrDevice", "log10::a"], [0, 1, 1, "_CPPv45log10RK5array14StreamOrDevice", "log10::s"], [0, 0, 1, "_CPPv45log1pRK5array14StreamOrDevice", "log1p"], [0, 1, 1, "_CPPv45log1pRK5array14StreamOrDevice", "log1p::a"], [0, 1, 1, "_CPPv45log1pRK5array14StreamOrDevice", "log1p::s"], [0, 0, 1, "_CPPv44log2RK5array14StreamOrDevice", "log2"], [0, 1, 1, "_CPPv44log2RK5array14StreamOrDevice", "log2::a"], [0, 1, 1, "_CPPv44log2RK5array14StreamOrDevice", "log2::s"], [0, 1, 1, "_CPPv43logRK5array14StreamOrDevice", "log::a"], [0, 1, 1, "_CPPv43logRK5array14StreamOrDevice", "log::s"], [0, 0, 1, "_CPPv49logaddexpRK5arrayRK5array14StreamOrDevice", "logaddexp"], [0, 1, 1, "_CPPv49logaddexpRK5arrayRK5array14StreamOrDevice", "logaddexp::a"], [0, 1, 1, "_CPPv49logaddexpRK5arrayRK5array14StreamOrDevice", "logaddexp::b"], [0, 1, 1, "_CPPv49logaddexpRK5arrayRK5array14StreamOrDevice", "logaddexp::s"], [0, 0, 1, "_CPPv411logical_andRK5arrayRK5array14StreamOrDevice", "logical_and"], [0, 1, 1, "_CPPv411logical_andRK5arrayRK5array14StreamOrDevice", "logical_and::a"], [0, 1, 1, "_CPPv411logical_andRK5arrayRK5array14StreamOrDevice", "logical_and::b"], [0, 1, 1, "_CPPv411logical_andRK5arrayRK5array14StreamOrDevice", "logical_and::s"], [0, 0, 1, "_CPPv411logical_notRK5array14StreamOrDevice", "logical_not"], [0, 1, 1, "_CPPv411logical_notRK5array14StreamOrDevice", "logical_not::a"], [0, 1, 1, "_CPPv411logical_notRK5array14StreamOrDevice", "logical_not::s"], [0, 0, 1, "_CPPv410logical_orRK5arrayRK5array14StreamOrDevice", "logical_or"], [0, 1, 1, "_CPPv410logical_orRK5arrayRK5array14StreamOrDevice", "logical_or::a"], [0, 1, 1, "_CPPv410logical_orRK5arrayRK5array14StreamOrDevice", "logical_or::b"], [0, 1, 1, "_CPPv410logical_orRK5arrayRK5array14StreamOrDevice", "logical_or::s"], [0, 0, 1, "_CPPv49logsumexpRK5array14StreamOrDevice", "logsumexp"], [0, 0, 1, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "logsumexp"], [0, 0, 1, "_CPPv49logsumexpRK5arrayb14StreamOrDevice", "logsumexp"], [0, 0, 1, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", "logsumexp"], [0, 1, 1, "_CPPv49logsumexpRK5array14StreamOrDevice", "logsumexp::a"], [0, 1, 1, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "logsumexp::a"], [0, 1, 1, "_CPPv49logsumexpRK5arrayb14StreamOrDevice", "logsumexp::a"], [0, 1, 1, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", "logsumexp::a"], [0, 1, 1, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "logsumexp::axes"], [0, 1, 1, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", "logsumexp::axis"], [0, 1, 1, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "logsumexp::keepdims"], [0, 1, 1, "_CPPv49logsumexpRK5arrayb14StreamOrDevice", "logsumexp::keepdims"], [0, 1, 1, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", "logsumexp::keepdims"], [0, 1, 1, "_CPPv49logsumexpRK5array14StreamOrDevice", "logsumexp::s"], [0, 1, 1, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "logsumexp::s"], [0, 1, 1, "_CPPv49logsumexpRK5arrayb14StreamOrDevice", "logsumexp::s"], [0, 1, 1, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", "logsumexp::s"], [0, 0, 1, "_CPPv46matmulRK5arrayRK5array14StreamOrDevice", "matmul"], [0, 1, 1, "_CPPv46matmulRK5arrayRK5array14StreamOrDevice", "matmul::a"], [0, 1, 1, "_CPPv46matmulRK5arrayRK5array14StreamOrDevice", "matmul::b"], [0, 1, 1, "_CPPv46matmulRK5arrayRK5array14StreamOrDevice", "matmul::s"], [0, 0, 1, "_CPPv43maxRK5array14StreamOrDevice", "max"], [0, 0, 1, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "max"], [0, 0, 1, "_CPPv43maxRK5arrayb14StreamOrDevice", "max"], [0, 0, 1, "_CPPv43maxRK5arrayib14StreamOrDevice", "max"], [0, 1, 1, "_CPPv43maxRK5array14StreamOrDevice", "max::a"], [0, 1, 1, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "max::a"], [0, 1, 1, "_CPPv43maxRK5arrayb14StreamOrDevice", "max::a"], [0, 1, 1, "_CPPv43maxRK5arrayib14StreamOrDevice", "max::a"], [0, 1, 1, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "max::axes"], [0, 1, 1, "_CPPv43maxRK5arrayib14StreamOrDevice", "max::axis"], [0, 1, 1, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "max::keepdims"], [0, 1, 1, "_CPPv43maxRK5arrayb14StreamOrDevice", "max::keepdims"], [0, 1, 1, "_CPPv43maxRK5arrayib14StreamOrDevice", "max::keepdims"], [0, 1, 1, "_CPPv43maxRK5array14StreamOrDevice", "max::s"], [0, 1, 1, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "max::s"], [0, 1, 1, "_CPPv43maxRK5arrayb14StreamOrDevice", "max::s"], [0, 1, 1, "_CPPv43maxRK5arrayib14StreamOrDevice", "max::s"], [0, 0, 1, "_CPPv47maximumRK5arrayRK5array14StreamOrDevice", "maximum"], [0, 1, 1, "_CPPv47maximumRK5arrayRK5array14StreamOrDevice", "maximum::a"], [0, 1, 1, "_CPPv47maximumRK5arrayRK5array14StreamOrDevice", "maximum::b"], [0, 1, 1, "_CPPv47maximumRK5arrayRK5array14StreamOrDevice", "maximum::s"], [0, 0, 1, "_CPPv44meanRK5array14StreamOrDevice", "mean"], [0, 0, 1, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "mean"], [0, 0, 1, "_CPPv44meanRK5arrayb14StreamOrDevice", "mean"], [0, 0, 1, "_CPPv44meanRK5arrayib14StreamOrDevice", "mean"], [0, 1, 1, "_CPPv44meanRK5array14StreamOrDevice", "mean::a"], [0, 1, 1, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "mean::a"], [0, 1, 1, "_CPPv44meanRK5arrayb14StreamOrDevice", "mean::a"], [0, 1, 1, "_CPPv44meanRK5arrayib14StreamOrDevice", "mean::a"], [0, 1, 1, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "mean::axes"], [0, 1, 1, "_CPPv44meanRK5arrayib14StreamOrDevice", "mean::axis"], [0, 1, 1, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "mean::keepdims"], [0, 1, 1, "_CPPv44meanRK5arrayb14StreamOrDevice", "mean::keepdims"], [0, 1, 1, "_CPPv44meanRK5arrayib14StreamOrDevice", "mean::keepdims"], [0, 1, 1, "_CPPv44meanRK5array14StreamOrDevice", "mean::s"], [0, 1, 1, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "mean::s"], [0, 1, 1, "_CPPv44meanRK5arrayb14StreamOrDevice", "mean::s"], [0, 1, 1, "_CPPv44meanRK5arrayib14StreamOrDevice", "mean::s"], [0, 0, 1, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", "meshgrid"], [0, 1, 1, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", "meshgrid::arrays"], [0, 1, 1, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", "meshgrid::indexing"], [0, 1, 1, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", "meshgrid::s"], [0, 1, 1, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", "meshgrid::sparse"], [0, 0, 1, "_CPPv43minRK5array14StreamOrDevice", "min"], [0, 0, 1, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "min"], [0, 0, 1, "_CPPv43minRK5arrayb14StreamOrDevice", "min"], [0, 0, 1, "_CPPv43minRK5arrayib14StreamOrDevice", "min"], [0, 1, 1, "_CPPv43minRK5array14StreamOrDevice", "min::a"], [0, 1, 1, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "min::a"], [0, 1, 1, "_CPPv43minRK5arrayb14StreamOrDevice", "min::a"], [0, 1, 1, "_CPPv43minRK5arrayib14StreamOrDevice", "min::a"], [0, 1, 1, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "min::axes"], [0, 1, 1, "_CPPv43minRK5arrayib14StreamOrDevice", "min::axis"], [0, 1, 1, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "min::keepdims"], [0, 1, 1, "_CPPv43minRK5arrayb14StreamOrDevice", "min::keepdims"], [0, 1, 1, "_CPPv43minRK5arrayib14StreamOrDevice", "min::keepdims"], [0, 1, 1, "_CPPv43minRK5array14StreamOrDevice", "min::s"], [0, 1, 1, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "min::s"], [0, 1, 1, "_CPPv43minRK5arrayb14StreamOrDevice", "min::s"], [0, 1, 1, "_CPPv43minRK5arrayib14StreamOrDevice", "min::s"], [0, 0, 1, "_CPPv47minimumRK5arrayRK5array14StreamOrDevice", "minimum"], [0, 1, 1, "_CPPv47minimumRK5arrayRK5array14StreamOrDevice", "minimum::a"], [0, 1, 1, "_CPPv47minimumRK5arrayRK5array14StreamOrDevice", "minimum::b"], [0, 1, 1, "_CPPv47minimumRK5arrayRK5array14StreamOrDevice", "minimum::s"], [0, 0, 1, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", "moveaxis"], [0, 1, 1, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", "moveaxis::a"], [0, 1, 1, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", "moveaxis::destination"], [0, 1, 1, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", "moveaxis::s"], [0, 1, 1, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", "moveaxis::source"], [0, 0, 1, "_CPPv48multiplyRK5arrayRK5array14StreamOrDevice", "multiply"], [0, 1, 1, "_CPPv48multiplyRK5arrayRK5array14StreamOrDevice", "multiply::a"], [0, 1, 1, "_CPPv48multiplyRK5arrayRK5array14StreamOrDevice", "multiply::b"], [0, 1, 1, "_CPPv48multiplyRK5arrayRK5array14StreamOrDevice", "multiply::s"], [0, 0, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num"], [0, 1, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num::a"], [0, 1, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num::nan"], [0, 1, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num::neginf"], [0, 1, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num::posinf"], [0, 1, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num::s"], [0, 0, 1, "_CPPv48negativeRK5array14StreamOrDevice", "negative"], [0, 1, 1, "_CPPv48negativeRK5array14StreamOrDevice", "negative::a"], [0, 1, 1, "_CPPv48negativeRK5array14StreamOrDevice", "negative::s"], [0, 0, 1, "_CPPv49not_equalRK5arrayRK5array14StreamOrDevice", "not_equal"], [0, 1, 1, "_CPPv49not_equalRK5arrayRK5array14StreamOrDevice", "not_equal::a"], [0, 1, 1, "_CPPv49not_equalRK5arrayRK5array14StreamOrDevice", "not_equal::b"], [0, 1, 1, "_CPPv49not_equalRK5arrayRK5array14StreamOrDevice", "not_equal::s"], [0, 0, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements"], [0, 1, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements::a"], [0, 1, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements::axes"], [0, 1, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements::dtype"], [0, 1, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements::inverted"], [0, 1, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements::s"], [0, 0, 1, "_CPPv44onesRKNSt6vectorIiEE14StreamOrDevice", "ones"], [0, 0, 1, "_CPPv44onesRKNSt6vectorIiEE5Dtype14StreamOrDevice", "ones"], [0, 1, 1, "_CPPv44onesRKNSt6vectorIiEE5Dtype14StreamOrDevice", "ones::dtype"], [0, 1, 1, "_CPPv44onesRKNSt6vectorIiEE14StreamOrDevice", "ones::s"], [0, 1, 1, "_CPPv44onesRKNSt6vectorIiEE5Dtype14StreamOrDevice", "ones::s"], [0, 1, 1, "_CPPv44onesRKNSt6vectorIiEE14StreamOrDevice", "ones::shape"], [0, 1, 1, "_CPPv44onesRKNSt6vectorIiEE5Dtype14StreamOrDevice", "ones::shape"], [0, 0, 1, "_CPPv49ones_likeRK5array14StreamOrDevice", "ones_like"], [0, 1, 1, "_CPPv49ones_likeRK5array14StreamOrDevice", "ones_like::a"], [0, 1, 1, "_CPPv49ones_likeRK5array14StreamOrDevice", "ones_like::s"], [0, 0, 1, "_CPPv4I0Ene5array1TRK5array", "operator!="], [0, 0, 1, "_CPPv4I0Ene5arrayRK5array1T", "operator!="], [0, 0, 1, "_CPPv4neRK5arrayRK5array", "operator!="], [0, 2, 1, "_CPPv4I0Ene5array1TRK5array", "operator!=::T"], [0, 2, 1, "_CPPv4I0Ene5arrayRK5array1T", "operator!=::T"], [0, 1, 1, "_CPPv4I0Ene5array1TRK5array", "operator!=::a"], [0, 1, 1, "_CPPv4I0Ene5arrayRK5array1T", "operator!=::a"], [0, 1, 1, "_CPPv4neRK5arrayRK5array", "operator!=::a"], [0, 1, 1, "_CPPv4I0Ene5array1TRK5array", "operator!=::b"], [0, 1, 1, "_CPPv4I0Ene5arrayRK5array1T", "operator!=::b"], [0, 1, 1, "_CPPv4neRK5arrayRK5array", "operator!=::b"], [0, 0, 1, "_CPPv4I0Erm5array1TRK5array", "operator%"], [0, 0, 1, "_CPPv4I0Erm5arrayRK5array1T", "operator%"], [0, 0, 1, "_CPPv4rmRK5arrayRK5array", "operator%"], [0, 2, 1, "_CPPv4I0Erm5array1TRK5array", "operator%::T"], [0, 2, 1, "_CPPv4I0Erm5arrayRK5array1T", "operator%::T"], [0, 1, 1, "_CPPv4I0Erm5array1TRK5array", "operator%::a"], [0, 1, 1, "_CPPv4I0Erm5arrayRK5array1T", "operator%::a"], [0, 1, 1, "_CPPv4rmRK5arrayRK5array", "operator%::a"], [0, 1, 1, "_CPPv4I0Erm5array1TRK5array", "operator%::b"], [0, 1, 1, "_CPPv4I0Erm5arrayRK5array1T", "operator%::b"], [0, 1, 1, "_CPPv4rmRK5arrayRK5array", "operator%::b"], [0, 0, 1, "_CPPv4anRK5arrayRK5array", "operator&"], [0, 0, 1, "_CPPv4aaRK5arrayRK5array", "operator&&"], [0, 1, 1, "_CPPv4aaRK5arrayRK5array", "operator&&::a"], [0, 1, 1, "_CPPv4aaRK5arrayRK5array", "operator&&::b"], [0, 1, 1, "_CPPv4anRK5arrayRK5array", "operator&::a"], [0, 1, 1, "_CPPv4anRK5arrayRK5array", "operator&::b"], [0, 0, 1, "_CPPv4I0Eml5array1TRK5array", "operator*"], [0, 0, 1, "_CPPv4I0Eml5arrayRK5array1T", "operator*"], [0, 0, 1, "_CPPv4mlRK5arrayRK5array", "operator*"], [0, 2, 1, "_CPPv4I0Eml5array1TRK5array", "operator*::T"], [0, 2, 1, "_CPPv4I0Eml5arrayRK5array1T", "operator*::T"], [0, 1, 1, "_CPPv4I0Eml5array1TRK5array", "operator*::a"], [0, 1, 1, "_CPPv4I0Eml5arrayRK5array1T", "operator*::a"], [0, 1, 1, "_CPPv4mlRK5arrayRK5array", "operator*::a"], [0, 1, 1, "_CPPv4I0Eml5array1TRK5array", "operator*::b"], [0, 1, 1, "_CPPv4I0Eml5arrayRK5array1T", "operator*::b"], [0, 1, 1, "_CPPv4mlRK5arrayRK5array", "operator*::b"], [0, 0, 1, "_CPPv4I0Epl5array1TRK5array", "operator+"], [0, 0, 1, "_CPPv4I0Epl5arrayRK5array1T", "operator+"], [0, 0, 1, "_CPPv4plRK5arrayRK5array", "operator+"], [0, 2, 1, "_CPPv4I0Epl5array1TRK5array", "operator+::T"], [0, 2, 1, "_CPPv4I0Epl5arrayRK5array1T", "operator+::T"], [0, 1, 1, "_CPPv4I0Epl5array1TRK5array", "operator+::a"], [0, 1, 1, "_CPPv4I0Epl5arrayRK5array1T", "operator+::a"], [0, 1, 1, "_CPPv4plRK5arrayRK5array", "operator+::a"], [0, 1, 1, "_CPPv4I0Epl5array1TRK5array", "operator+::b"], [0, 1, 1, "_CPPv4I0Epl5arrayRK5array1T", "operator+::b"], [0, 1, 1, "_CPPv4plRK5arrayRK5array", "operator+::b"], [0, 0, 1, "_CPPv4I0Emi5array1TRK5array", "operator-"], [0, 0, 1, "_CPPv4I0Emi5arrayRK5array1T", "operator-"], [0, 0, 1, "_CPPv4miRK5array", "operator-"], [0, 0, 1, "_CPPv4miRK5arrayRK5array", "operator-"], [0, 2, 1, "_CPPv4I0Emi5array1TRK5array", "operator-::T"], [0, 2, 1, "_CPPv4I0Emi5arrayRK5array1T", "operator-::T"], [0, 1, 1, "_CPPv4I0Emi5array1TRK5array", "operator-::a"], [0, 1, 1, "_CPPv4I0Emi5arrayRK5array1T", "operator-::a"], [0, 1, 1, "_CPPv4miRK5array", "operator-::a"], [0, 1, 1, "_CPPv4miRK5arrayRK5array", "operator-::a"], [0, 1, 1, "_CPPv4I0Emi5array1TRK5array", "operator-::b"], [0, 1, 1, "_CPPv4I0Emi5arrayRK5array1T", "operator-::b"], [0, 1, 1, "_CPPv4miRK5arrayRK5array", "operator-::b"], [0, 0, 1, "_CPPv4dvRK5arrayRK5array", "operator/"], [0, 0, 1, "_CPPv4dvRK5arrayd", "operator/"], [0, 0, 1, "_CPPv4dvdRK5array", "operator/"], [0, 1, 1, "_CPPv4dvRK5arrayRK5array", "operator/::a"], [0, 1, 1, "_CPPv4dvRK5arrayd", "operator/::a"], [0, 1, 1, "_CPPv4dvdRK5array", "operator/::a"], [0, 1, 1, "_CPPv4dvRK5arrayRK5array", "operator/::b"], [0, 1, 1, "_CPPv4dvRK5arrayd", "operator/::b"], [0, 1, 1, "_CPPv4dvdRK5array", "operator/::b"], [0, 0, 1, "_CPPv4I0Elt5array1TRK5array", "operator<"], [0, 0, 1, "_CPPv4I0Elt5arrayRK5array1T", "operator<"], [0, 0, 1, "_CPPv4ltRK5arrayRK5array", "operator<"], [0, 2, 1, "_CPPv4I0Elt5array1TRK5array", "operator<::T"], [0, 2, 1, "_CPPv4I0Elt5arrayRK5array1T", "operator<::T"], [0, 1, 1, "_CPPv4I0Elt5array1TRK5array", "operator<::a"], [0, 1, 1, "_CPPv4I0Elt5arrayRK5array1T", "operator<::a"], [0, 1, 1, "_CPPv4ltRK5arrayRK5array", "operator<::a"], [0, 1, 1, "_CPPv4I0Elt5array1TRK5array", "operator<::b"], [0, 1, 1, "_CPPv4I0Elt5arrayRK5array1T", "operator<::b"], [0, 1, 1, "_CPPv4ltRK5arrayRK5array", "operator<::b"], [0, 0, 1, "_CPPv4lsRK5arrayRK5array", "operator<<"], [0, 1, 1, "_CPPv4lsRK5arrayRK5array", "operator<<::a"], [0, 1, 1, "_CPPv4lsRK5arrayRK5array", "operator<<::b"], [0, 0, 1, "_CPPv4I0Ele5array1TRK5array", "operator<="], [0, 0, 1, "_CPPv4I0Ele5arrayRK5array1T", "operator<="], [0, 0, 1, "_CPPv4leRK5arrayRK5array", "operator<="], [0, 2, 1, "_CPPv4I0Ele5array1TRK5array", "operator<=::T"], [0, 2, 1, "_CPPv4I0Ele5arrayRK5array1T", "operator<=::T"], [0, 1, 1, "_CPPv4I0Ele5array1TRK5array", "operator<=::a"], [0, 1, 1, "_CPPv4I0Ele5arrayRK5array1T", "operator<=::a"], [0, 1, 1, "_CPPv4leRK5arrayRK5array", "operator<=::a"], [0, 1, 1, "_CPPv4I0Ele5array1TRK5array", "operator<=::b"], [0, 1, 1, "_CPPv4I0Ele5arrayRK5array1T", "operator<=::b"], [0, 1, 1, "_CPPv4leRK5arrayRK5array", "operator<=::b"], [0, 0, 1, "_CPPv4I0Eeq5array1TRK5array", "operator=="], [0, 0, 1, "_CPPv4I0Eeq5arrayRK5array1T", "operator=="], [0, 0, 1, "_CPPv4eqRK5arrayRK5array", "operator=="], [0, 2, 1, "_CPPv4I0Eeq5array1TRK5array", "operator==::T"], [0, 2, 1, "_CPPv4I0Eeq5arrayRK5array1T", "operator==::T"], [0, 1, 1, "_CPPv4I0Eeq5array1TRK5array", "operator==::a"], [0, 1, 1, "_CPPv4I0Eeq5arrayRK5array1T", "operator==::a"], [0, 1, 1, "_CPPv4eqRK5arrayRK5array", "operator==::a"], [0, 1, 1, "_CPPv4I0Eeq5array1TRK5array", "operator==::b"], [0, 1, 1, "_CPPv4I0Eeq5arrayRK5array1T", "operator==::b"], [0, 1, 1, "_CPPv4eqRK5arrayRK5array", "operator==::b"], [0, 0, 1, "_CPPv4I0Egt5array1TRK5array", "operator>"], [0, 0, 1, "_CPPv4I0Egt5arrayRK5array1T", "operator>"], [0, 0, 1, "_CPPv4gtRK5arrayRK5array", "operator>"], [0, 2, 1, "_CPPv4I0Egt5array1TRK5array", "operator>::T"], [0, 2, 1, "_CPPv4I0Egt5arrayRK5array1T", "operator>::T"], [0, 1, 1, "_CPPv4I0Egt5array1TRK5array", "operator>::a"], [0, 1, 1, "_CPPv4I0Egt5arrayRK5array1T", "operator>::a"], [0, 1, 1, "_CPPv4gtRK5arrayRK5array", "operator>::a"], [0, 1, 1, "_CPPv4I0Egt5array1TRK5array", "operator>::b"], [0, 1, 1, "_CPPv4I0Egt5arrayRK5array1T", "operator>::b"], [0, 1, 1, "_CPPv4gtRK5arrayRK5array", "operator>::b"], [0, 0, 1, "_CPPv4I0Ege5array1TRK5array", "operator>="], [0, 0, 1, "_CPPv4I0Ege5arrayRK5array1T", "operator>="], [0, 0, 1, "_CPPv4geRK5arrayRK5array", "operator>="], [0, 2, 1, "_CPPv4I0Ege5array1TRK5array", "operator>=::T"], [0, 2, 1, "_CPPv4I0Ege5arrayRK5array1T", "operator>=::T"], [0, 1, 1, "_CPPv4I0Ege5array1TRK5array", "operator>=::a"], [0, 1, 1, "_CPPv4I0Ege5arrayRK5array1T", "operator>=::a"], [0, 1, 1, "_CPPv4geRK5arrayRK5array", "operator>=::a"], [0, 1, 1, "_CPPv4I0Ege5array1TRK5array", "operator>=::b"], [0, 1, 1, "_CPPv4I0Ege5arrayRK5array1T", "operator>=::b"], [0, 1, 1, "_CPPv4geRK5arrayRK5array", "operator>=::b"], [0, 0, 1, "_CPPv4rsRK5arrayRK5array", "operator>>"], [0, 1, 1, "_CPPv4rsRK5arrayRK5array", "operator>>::a"], [0, 1, 1, "_CPPv4rsRK5arrayRK5array", "operator>>::b"], [0, 0, 1, "_CPPv4eoRK5arrayRK5array", "operator^"], [0, 1, 1, "_CPPv4eoRK5arrayRK5array", "operator^::a"], [0, 1, 1, "_CPPv4eoRK5arrayRK5array", "operator^::b"], [0, 0, 1, "_CPPv4orRK5arrayRK5array", "operator|"], [0, 1, 1, "_CPPv4orRK5arrayRK5array", "operator|::a"], [0, 1, 1, "_CPPv4orRK5arrayRK5array", "operator|::b"], [0, 0, 1, "_CPPv4ooRK5arrayRK5array", "operator||"], [0, 1, 1, "_CPPv4ooRK5arrayRK5array", "operator||::a"], [0, 1, 1, "_CPPv4ooRK5arrayRK5array", "operator||::b"], [0, 0, 1, "_CPPv45outerRK5arrayRK5array14StreamOrDevice", "outer"], [0, 1, 1, "_CPPv45outerRK5arrayRK5array14StreamOrDevice", "outer::a"], [0, 1, 1, "_CPPv45outerRK5arrayRK5array14StreamOrDevice", "outer::b"], [0, 1, 1, "_CPPv45outerRK5arrayRK5array14StreamOrDevice", "outer::s"], [0, 0, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad"], [0, 0, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad"], [0, 0, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad"], [0, 0, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::a"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad::a"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::a"], [0, 1, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad::a"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::axes"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::high_pad_size"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::low_pad_size"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::mode"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad::mode"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::mode"], [0, 1, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad::mode"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_value"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_value"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_value"], [0, 1, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_value"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_width"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_width"], [0, 1, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_width"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::s"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad::s"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::s"], [0, 1, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad::s"], [0, 0, 1, "_CPPv49partitionRK5arrayi14StreamOrDevice", "partition"], [0, 0, 1, "_CPPv49partitionRK5arrayii14StreamOrDevice", "partition"], [0, 1, 1, "_CPPv49partitionRK5arrayi14StreamOrDevice", "partition::a"], [0, 1, 1, "_CPPv49partitionRK5arrayii14StreamOrDevice", "partition::a"], [0, 1, 1, "_CPPv49partitionRK5arrayii14StreamOrDevice", "partition::axis"], [0, 1, 1, "_CPPv49partitionRK5arrayi14StreamOrDevice", "partition::kth"], [0, 1, 1, "_CPPv49partitionRK5arrayii14StreamOrDevice", "partition::kth"], [0, 1, 1, "_CPPv49partitionRK5arrayi14StreamOrDevice", "partition::s"], [0, 1, 1, "_CPPv49partitionRK5arrayii14StreamOrDevice", "partition::s"], [0, 0, 1, "_CPPv45powerRK5arrayRK5array14StreamOrDevice", "power"], [0, 1, 1, "_CPPv45powerRK5arrayRK5array14StreamOrDevice", "power::a"], [0, 1, 1, "_CPPv45powerRK5arrayRK5array14StreamOrDevice", "power::b"], [0, 1, 1, "_CPPv45powerRK5arrayRK5array14StreamOrDevice", "power::s"], [0, 0, 1, "_CPPv44prodRK5array14StreamOrDevice", "prod"], [0, 0, 1, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "prod"], [0, 0, 1, "_CPPv44prodRK5arrayb14StreamOrDevice", "prod"], [0, 0, 1, "_CPPv44prodRK5arrayib14StreamOrDevice", "prod"], [0, 1, 1, "_CPPv44prodRK5array14StreamOrDevice", "prod::a"], [0, 1, 1, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "prod::a"], [0, 1, 1, "_CPPv44prodRK5arrayb14StreamOrDevice", "prod::a"], [0, 1, 1, "_CPPv44prodRK5arrayib14StreamOrDevice", "prod::a"], [0, 1, 1, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "prod::axes"], [0, 1, 1, "_CPPv44prodRK5arrayib14StreamOrDevice", "prod::axis"], [0, 1, 1, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "prod::keepdims"], [0, 1, 1, "_CPPv44prodRK5arrayb14StreamOrDevice", "prod::keepdims"], [0, 1, 1, "_CPPv44prodRK5arrayib14StreamOrDevice", "prod::keepdims"], [0, 1, 1, "_CPPv44prodRK5array14StreamOrDevice", "prod::s"], [0, 1, 1, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "prod::s"], [0, 1, 1, "_CPPv44prodRK5arrayb14StreamOrDevice", "prod::s"], [0, 1, 1, "_CPPv44prodRK5arrayib14StreamOrDevice", "prod::s"], [0, 0, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis"], [0, 1, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis::a"], [0, 1, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis::axis"], [0, 1, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis::indices"], [0, 1, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis::s"], [0, 1, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis::values"], [0, 0, 1, "_CPPv48quantizeRK5arrayii14StreamOrDevice", "quantize"], [0, 1, 1, "_CPPv48quantizeRK5arrayii14StreamOrDevice", "quantize::bits"], [0, 1, 1, "_CPPv48quantizeRK5arrayii14StreamOrDevice", "quantize::group_size"], [0, 1, 1, "_CPPv48quantizeRK5arrayii14StreamOrDevice", "quantize::s"], [0, 1, 1, "_CPPv48quantizeRK5arrayii14StreamOrDevice", "quantize::w"], [0, 0, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::biases"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::bits"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::group_size"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::s"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::scales"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::transpose"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::w"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::x"], [0, 0, 1, "_CPPv47radiansRK5array14StreamOrDevice", "radians"], [0, 1, 1, "_CPPv47radiansRK5array14StreamOrDevice", "radians::a"], [0, 1, 1, "_CPPv47radiansRK5array14StreamOrDevice", "radians::s"], [0, 0, 1, "_CPPv44realRK5array14StreamOrDevice", "real"], [0, 1, 1, "_CPPv44realRK5array14StreamOrDevice", "real::a"], [0, 1, 1, "_CPPv44realRK5array14StreamOrDevice", "real::s"], [0, 0, 1, "_CPPv410reciprocalRK5array14StreamOrDevice", "reciprocal"], [0, 1, 1, "_CPPv410reciprocalRK5array14StreamOrDevice", "reciprocal::a"], [0, 1, 1, "_CPPv410reciprocalRK5array14StreamOrDevice", "reciprocal::s"], [0, 0, 1, "_CPPv49remainderRK5arrayRK5array14StreamOrDevice", "remainder"], [0, 1, 1, "_CPPv49remainderRK5arrayRK5array14StreamOrDevice", "remainder::a"], [0, 1, 1, "_CPPv49remainderRK5arrayRK5array14StreamOrDevice", "remainder::b"], [0, 1, 1, "_CPPv49remainderRK5arrayRK5array14StreamOrDevice", "remainder::s"], [0, 0, 1, "_CPPv46repeatRK5arrayi14StreamOrDevice", "repeat"], [0, 0, 1, "_CPPv46repeatRK5arrayii14StreamOrDevice", "repeat"], [0, 1, 1, "_CPPv46repeatRK5arrayi14StreamOrDevice", "repeat::arr"], [0, 1, 1, "_CPPv46repeatRK5arrayii14StreamOrDevice", "repeat::arr"], [0, 1, 1, "_CPPv46repeatRK5arrayii14StreamOrDevice", "repeat::axis"], [0, 1, 1, "_CPPv46repeatRK5arrayi14StreamOrDevice", "repeat::repeats"], [0, 1, 1, "_CPPv46repeatRK5arrayii14StreamOrDevice", "repeat::repeats"], [0, 1, 1, "_CPPv46repeatRK5arrayi14StreamOrDevice", "repeat::s"], [0, 1, 1, "_CPPv46repeatRK5arrayii14StreamOrDevice", "repeat::s"], [0, 0, 1, "_CPPv47reshapeRK5arrayNSt6vectorIiEE14StreamOrDevice", "reshape"], [0, 1, 1, "_CPPv47reshapeRK5arrayNSt6vectorIiEE14StreamOrDevice", "reshape::a"], [0, 1, 1, "_CPPv47reshapeRK5arrayNSt6vectorIiEE14StreamOrDevice", "reshape::s"], [0, 1, 1, "_CPPv47reshapeRK5arrayNSt6vectorIiEE14StreamOrDevice", "reshape::shape"], [0, 0, 1, "_CPPv411right_shiftRK5arrayRK5array14StreamOrDevice", "right_shift"], [0, 1, 1, "_CPPv411right_shiftRK5arrayRK5array14StreamOrDevice", "right_shift::a"], [0, 1, 1, "_CPPv411right_shiftRK5arrayRK5array14StreamOrDevice", "right_shift::b"], [0, 1, 1, "_CPPv411right_shiftRK5arrayRK5array14StreamOrDevice", "right_shift::s"], [0, 0, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "roll"], [0, 0, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "roll"], [0, 0, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "roll"], [0, 0, 1, "_CPPv44rollRK5arrayi14StreamOrDevice", "roll"], [0, 0, 1, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "roll"], [0, 0, 1, "_CPPv44rollRK5arrayii14StreamOrDevice", "roll"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayi14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayii14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "roll::axes"], [0, 1, 1, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "roll::axes"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "roll::axis"], [0, 1, 1, "_CPPv44rollRK5arrayii14StreamOrDevice", "roll::axis"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayi14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayii14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "roll::shift"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "roll::shift"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "roll::shift"], [0, 1, 1, "_CPPv44rollRK5arrayi14StreamOrDevice", "roll::shift"], [0, 1, 1, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "roll::shift"], [0, 1, 1, "_CPPv44rollRK5arrayii14StreamOrDevice", "roll::shift"], [0, 0, 1, "_CPPv45roundRK5array14StreamOrDevice", "round"], [0, 0, 1, "_CPPv45roundRK5arrayi14StreamOrDevice", "round"], [0, 1, 1, "_CPPv45roundRK5array14StreamOrDevice", "round::a"], [0, 1, 1, "_CPPv45roundRK5arrayi14StreamOrDevice", "round::a"], [0, 1, 1, "_CPPv45roundRK5arrayi14StreamOrDevice", "round::decimals"], [0, 1, 1, "_CPPv45roundRK5array14StreamOrDevice", "round::s"], [0, 1, 1, "_CPPv45roundRK5arrayi14StreamOrDevice", "round::s"], [0, 0, 1, "_CPPv45rsqrtRK5array14StreamOrDevice", "rsqrt"], [0, 1, 1, "_CPPv45rsqrtRK5array14StreamOrDevice", "rsqrt::a"], [0, 1, 1, "_CPPv45rsqrtRK5array14StreamOrDevice", "rsqrt::s"], [0, 0, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter"], [0, 0, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter"], [0, 1, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter::a"], [0, 1, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter::a"], [0, 1, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter::axes"], [0, 1, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter::axis"], [0, 1, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter::indices"], [0, 1, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter::indices"], [0, 1, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter::s"], [0, 1, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter::s"], [0, 1, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter::updates"], [0, 1, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter::updates"], [0, 0, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add"], [0, 0, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add::a"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add::a"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add::axes"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add::axis"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add::indices"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add::indices"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add::s"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add::s"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add::updates"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add::updates"], [0, 0, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max"], [0, 0, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max::a"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max::a"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max::axes"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max::axis"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max::indices"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max::indices"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max::s"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max::s"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max::updates"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max::updates"], [0, 0, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min"], [0, 0, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min::a"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min::a"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min::axes"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min::axis"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min::indices"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min::indices"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min::s"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min::s"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min::updates"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min::updates"], [0, 0, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod"], [0, 0, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod::a"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod::a"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod::axes"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod::axis"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod::indices"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod::indices"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod::s"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod::s"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod::updates"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod::updates"], [0, 0, 1, "_CPPv47sigmoidRK5array14StreamOrDevice", "sigmoid"], [0, 1, 1, "_CPPv47sigmoidRK5array14StreamOrDevice", "sigmoid::a"], [0, 1, 1, "_CPPv47sigmoidRK5array14StreamOrDevice", "sigmoid::s"], [0, 0, 1, "_CPPv44signRK5array14StreamOrDevice", "sign"], [0, 1, 1, "_CPPv44signRK5array14StreamOrDevice", "sign::a"], [0, 1, 1, "_CPPv44signRK5array14StreamOrDevice", "sign::s"], [0, 0, 1, "_CPPv43sinRK5array14StreamOrDevice", "sin"], [0, 1, 1, "_CPPv43sinRK5array14StreamOrDevice", "sin::a"], [0, 1, 1, "_CPPv43sinRK5array14StreamOrDevice", "sin::s"], [0, 0, 1, "_CPPv44sinhRK5array14StreamOrDevice", "sinh"], [0, 1, 1, "_CPPv44sinhRK5array14StreamOrDevice", "sinh::a"], [0, 1, 1, "_CPPv44sinhRK5array14StreamOrDevice", "sinh::s"], [0, 0, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice"], [0, 0, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::a"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::a"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::s"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::s"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::start"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::start"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::stop"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::stop"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::strides"], [0, 0, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update"], [0, 0, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::s"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::s"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::src"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::src"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::start"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::start"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::stop"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::stop"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::strides"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::update"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::update"], [0, 0, 1, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "softmax"], [0, 0, 1, "_CPPv47softmaxRK5arrayb14StreamOrDevice", "softmax"], [0, 0, 1, "_CPPv47softmaxRK5arrayib14StreamOrDevice", "softmax"], [0, 1, 1, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "softmax::a"], [0, 1, 1, "_CPPv47softmaxRK5arrayb14StreamOrDevice", "softmax::a"], [0, 1, 1, "_CPPv47softmaxRK5arrayib14StreamOrDevice", "softmax::a"], [0, 1, 1, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "softmax::axes"], [0, 1, 1, "_CPPv47softmaxRK5arrayib14StreamOrDevice", "softmax::axis"], [0, 1, 1, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "softmax::precise"], [0, 1, 1, "_CPPv47softmaxRK5arrayb14StreamOrDevice", "softmax::precise"], [0, 1, 1, "_CPPv47softmaxRK5arrayib14StreamOrDevice", "softmax::precise"], [0, 1, 1, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "softmax::s"], [0, 1, 1, "_CPPv47softmaxRK5arrayb14StreamOrDevice", "softmax::s"], [0, 1, 1, "_CPPv47softmaxRK5arrayib14StreamOrDevice", "softmax::s"], [0, 0, 1, "_CPPv44sortRK5array14StreamOrDevice", "sort"], [0, 0, 1, "_CPPv44sortRK5arrayi14StreamOrDevice", "sort"], [0, 1, 1, "_CPPv44sortRK5array14StreamOrDevice", "sort::a"], [0, 1, 1, "_CPPv44sortRK5arrayi14StreamOrDevice", "sort::a"], [0, 1, 1, "_CPPv44sortRK5arrayi14StreamOrDevice", "sort::axis"], [0, 1, 1, "_CPPv44sortRK5array14StreamOrDevice", "sort::s"], [0, 1, 1, "_CPPv44sortRK5arrayi14StreamOrDevice", "sort::s"], [0, 0, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "split"], [0, 0, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "split"], [0, 0, 1, "_CPPv45splitRK5arrayi14StreamOrDevice", "split"], [0, 0, 1, "_CPPv45splitRK5arrayii14StreamOrDevice", "split"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "split::a"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "split::a"], [0, 1, 1, "_CPPv45splitRK5arrayi14StreamOrDevice", "split::a"], [0, 1, 1, "_CPPv45splitRK5arrayii14StreamOrDevice", "split::a"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "split::axis"], [0, 1, 1, "_CPPv45splitRK5arrayii14StreamOrDevice", "split::axis"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "split::indices"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "split::indices"], [0, 1, 1, "_CPPv45splitRK5arrayi14StreamOrDevice", "split::num_splits"], [0, 1, 1, "_CPPv45splitRK5arrayii14StreamOrDevice", "split::num_splits"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "split::s"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "split::s"], [0, 1, 1, "_CPPv45splitRK5arrayi14StreamOrDevice", "split::s"], [0, 1, 1, "_CPPv45splitRK5arrayii14StreamOrDevice", "split::s"], [0, 0, 1, "_CPPv44sqrtRK5array14StreamOrDevice", "sqrt"], [0, 1, 1, "_CPPv44sqrtRK5array14StreamOrDevice", "sqrt::a"], [0, 1, 1, "_CPPv44sqrtRK5array14StreamOrDevice", "sqrt::s"], [0, 0, 1, "_CPPv46squareRK5array14StreamOrDevice", "square"], [0, 1, 1, "_CPPv46squareRK5array14StreamOrDevice", "square::a"], [0, 1, 1, "_CPPv46squareRK5array14StreamOrDevice", "square::s"], [0, 0, 1, "_CPPv47squeezeRK5array14StreamOrDevice", "squeeze"], [0, 0, 1, "_CPPv47squeezeRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "squeeze"], [0, 0, 1, "_CPPv47squeezeRK5arrayi14StreamOrDevice", "squeeze"], [0, 1, 1, "_CPPv47squeezeRK5array14StreamOrDevice", "squeeze::a"], [0, 1, 1, "_CPPv47squeezeRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "squeeze::a"], [0, 1, 1, "_CPPv47squeezeRK5arrayi14StreamOrDevice", "squeeze::a"], [0, 1, 1, "_CPPv47squeezeRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "squeeze::axes"], [0, 1, 1, "_CPPv47squeezeRK5arrayi14StreamOrDevice", "squeeze::axis"], [0, 1, 1, "_CPPv47squeezeRK5array14StreamOrDevice", "squeeze::s"], [0, 1, 1, "_CPPv47squeezeRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "squeeze::s"], [0, 1, 1, "_CPPv47squeezeRK5arrayi14StreamOrDevice", "squeeze::s"], [0, 0, 1, "_CPPv45stackRKNSt6vectorI5arrayEE14StreamOrDevice", "stack"], [0, 0, 1, "_CPPv45stackRKNSt6vectorI5arrayEEi14StreamOrDevice", "stack"], [0, 1, 1, "_CPPv45stackRKNSt6vectorI5arrayEE14StreamOrDevice", "stack::arrays"], [0, 1, 1, "_CPPv45stackRKNSt6vectorI5arrayEEi14StreamOrDevice", "stack::arrays"], [0, 1, 1, "_CPPv45stackRKNSt6vectorI5arrayEEi14StreamOrDevice", "stack::axis"], [0, 1, 1, "_CPPv45stackRKNSt6vectorI5arrayEE14StreamOrDevice", "stack::s"], [0, 1, 1, "_CPPv45stackRKNSt6vectorI5arrayEEi14StreamOrDevice", "stack::s"], [0, 0, 1, "_CPPv4StRK5array14StreamOrDevice", "std"], [0, 0, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std"], [0, 0, 1, "_CPPv4StRK5arraybi14StreamOrDevice", "std"], [0, 0, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std"], [0, 1, 1, "_CPPv4StRK5array14StreamOrDevice", "std::a"], [0, 1, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std::a"], [0, 1, 1, "_CPPv4StRK5arraybi14StreamOrDevice", "std::a"], [0, 1, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std::a"], [0, 1, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std::axes"], [0, 1, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std::axis"], [0, 1, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std::ddof"], [0, 1, 1, "_CPPv4StRK5arraybi14StreamOrDevice", "std::ddof"], [0, 1, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std::ddof"], [0, 1, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std::keepdims"], [0, 1, 1, "_CPPv4StRK5arraybi14StreamOrDevice", "std::keepdims"], [0, 1, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std::keepdims"], [0, 1, 1, "_CPPv4StRK5array14StreamOrDevice", "std::s"], [0, 1, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std::s"], [0, 1, 1, "_CPPv4StRK5arraybi14StreamOrDevice", "std::s"], [0, 1, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std::s"], [0, 0, 1, "_CPPv413stop_gradientRK5array14StreamOrDevice", "stop_gradient"], [0, 1, 1, "_CPPv413stop_gradientRK5array14StreamOrDevice", "stop_gradient::a"], [0, 1, 1, "_CPPv413stop_gradientRK5array14StreamOrDevice", "stop_gradient::s"], [0, 0, 1, "_CPPv48subtractRK5arrayRK5array14StreamOrDevice", "subtract"], [0, 1, 1, "_CPPv48subtractRK5arrayRK5array14StreamOrDevice", "subtract::a"], [0, 1, 1, "_CPPv48subtractRK5arrayRK5array14StreamOrDevice", "subtract::b"], [0, 1, 1, "_CPPv48subtractRK5arrayRK5array14StreamOrDevice", "subtract::s"], [0, 0, 1, "_CPPv43sumRK5array14StreamOrDevice", "sum"], [0, 0, 1, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "sum"], [0, 0, 1, "_CPPv43sumRK5arrayb14StreamOrDevice", "sum"], [0, 0, 1, "_CPPv43sumRK5arrayib14StreamOrDevice", "sum"], [0, 1, 1, "_CPPv43sumRK5array14StreamOrDevice", "sum::a"], [0, 1, 1, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "sum::a"], [0, 1, 1, "_CPPv43sumRK5arrayb14StreamOrDevice", "sum::a"], [0, 1, 1, "_CPPv43sumRK5arrayib14StreamOrDevice", "sum::a"], [0, 1, 1, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "sum::axes"], [0, 1, 1, "_CPPv43sumRK5arrayib14StreamOrDevice", "sum::axis"], [0, 1, 1, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "sum::keepdims"], [0, 1, 1, "_CPPv43sumRK5arrayb14StreamOrDevice", "sum::keepdims"], [0, 1, 1, "_CPPv43sumRK5arrayib14StreamOrDevice", "sum::keepdims"], [0, 1, 1, "_CPPv43sumRK5array14StreamOrDevice", "sum::s"], [0, 1, 1, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "sum::s"], [0, 1, 1, "_CPPv43sumRK5arrayb14StreamOrDevice", "sum::s"], [0, 1, 1, "_CPPv43sumRK5arrayib14StreamOrDevice", "sum::s"], [0, 0, 1, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", "swapaxes"], [0, 1, 1, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", "swapaxes::a"], [0, 1, 1, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", "swapaxes::axis1"], [0, 1, 1, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", "swapaxes::axis2"], [0, 1, 1, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", "swapaxes::s"], [0, 0, 1, "_CPPv44takeRK5arrayRK5array14StreamOrDevice", "take"], [0, 0, 1, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", "take"], [0, 0, 1, "_CPPv44takeRK5arrayi14StreamOrDevice", "take"], [0, 0, 1, "_CPPv44takeRK5arrayii14StreamOrDevice", "take"], [0, 1, 1, "_CPPv44takeRK5arrayRK5array14StreamOrDevice", "take::a"], [0, 1, 1, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", "take::a"], [0, 1, 1, "_CPPv44takeRK5arrayi14StreamOrDevice", "take::a"], [0, 1, 1, "_CPPv44takeRK5arrayii14StreamOrDevice", "take::a"], [0, 1, 1, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", "take::axis"], [0, 1, 1, "_CPPv44takeRK5arrayii14StreamOrDevice", "take::axis"], [0, 1, 1, "_CPPv44takeRK5arrayi14StreamOrDevice", "take::index"], [0, 1, 1, "_CPPv44takeRK5arrayii14StreamOrDevice", "take::index"], [0, 1, 1, "_CPPv44takeRK5arrayRK5array14StreamOrDevice", "take::indices"], [0, 1, 1, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", "take::indices"], [0, 1, 1, "_CPPv44takeRK5arrayRK5array14StreamOrDevice", "take::s"], [0, 1, 1, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", "take::s"], [0, 1, 1, "_CPPv44takeRK5arrayi14StreamOrDevice", "take::s"], [0, 1, 1, "_CPPv44takeRK5arrayii14StreamOrDevice", "take::s"], [0, 0, 1, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", "take_along_axis"], [0, 1, 1, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", "take_along_axis::a"], [0, 1, 1, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", "take_along_axis::axis"], [0, 1, 1, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", "take_along_axis::indices"], [0, 1, 1, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", "take_along_axis::s"], [0, 0, 1, "_CPPv43tanRK5array14StreamOrDevice", "tan"], [0, 1, 1, "_CPPv43tanRK5array14StreamOrDevice", "tan::a"], [0, 1, 1, "_CPPv43tanRK5array14StreamOrDevice", "tan::s"], [0, 0, 1, "_CPPv44tanhRK5array14StreamOrDevice", "tanh"], [0, 1, 1, "_CPPv44tanhRK5array14StreamOrDevice", "tanh::a"], [0, 1, 1, "_CPPv44tanhRK5array14StreamOrDevice", "tanh::s"], [0, 0, 1, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", "tensordot"], [0, 0, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", "tensordot::a"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot::a"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot::axes_a"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot::axes_b"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", "tensordot::axis"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", "tensordot::b"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot::b"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", "tensordot::s"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot::s"], [0, 0, 1, "_CPPv44tileRK5arrayNSt6vectorIiEE14StreamOrDevice", "tile"], [0, 1, 1, "_CPPv44tileRK5arrayNSt6vectorIiEE14StreamOrDevice", "tile::arr"], [0, 1, 1, "_CPPv44tileRK5arrayNSt6vectorIiEE14StreamOrDevice", "tile::reps"], [0, 1, 1, "_CPPv44tileRK5arrayNSt6vectorIiEE14StreamOrDevice", "tile::s"], [0, 0, 1, "_CPPv44topkRK5arrayi14StreamOrDevice", "topk"], [0, 0, 1, "_CPPv44topkRK5arrayii14StreamOrDevice", "topk"], [0, 1, 1, "_CPPv44topkRK5arrayi14StreamOrDevice", "topk::a"], [0, 1, 1, "_CPPv44topkRK5arrayii14StreamOrDevice", "topk::a"], [0, 1, 1, "_CPPv44topkRK5arrayii14StreamOrDevice", "topk::axis"], [0, 1, 1, "_CPPv44topkRK5arrayi14StreamOrDevice", "topk::k"], [0, 1, 1, "_CPPv44topkRK5arrayii14StreamOrDevice", "topk::k"], [0, 1, 1, "_CPPv44topkRK5arrayi14StreamOrDevice", "topk::s"], [0, 1, 1, "_CPPv44topkRK5arrayii14StreamOrDevice", "topk::s"], [0, 0, 1, "_CPPv45traceRK5array14StreamOrDevice", "trace"], [0, 0, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace"], [0, 0, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace"], [0, 1, 1, "_CPPv45traceRK5array14StreamOrDevice", "trace::a"], [0, 1, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace::a"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::a"], [0, 1, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace::axis1"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::axis1"], [0, 1, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace::axis2"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::axis2"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::dtype"], [0, 1, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace::offset"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::offset"], [0, 1, 1, "_CPPv45traceRK5array14StreamOrDevice", "trace::s"], [0, 1, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace::s"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::s"], [0, 0, 1, "_CPPv49transposeRK5array14StreamOrDevice", "transpose"], [0, 0, 1, "_CPPv49transposeRK5arrayNSt16initializer_listIiEE14StreamOrDevice", "transpose"], [0, 0, 1, "_CPPv49transposeRK5arrayNSt6vectorIiEE14StreamOrDevice", "transpose"], [0, 1, 1, "_CPPv49transposeRK5array14StreamOrDevice", "transpose::a"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt16initializer_listIiEE14StreamOrDevice", "transpose::a"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt6vectorIiEE14StreamOrDevice", "transpose::a"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt16initializer_listIiEE14StreamOrDevice", "transpose::axes"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt6vectorIiEE14StreamOrDevice", "transpose::axes"], [0, 1, 1, "_CPPv49transposeRK5array14StreamOrDevice", "transpose::s"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt16initializer_listIiEE14StreamOrDevice", "transpose::s"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt6vectorIiEE14StreamOrDevice", "transpose::s"], [0, 0, 1, "_CPPv43trii5Dtype14StreamOrDevice", "tri"], [0, 0, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri"], [0, 1, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri::k"], [0, 1, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri::m"], [0, 1, 1, "_CPPv43trii5Dtype14StreamOrDevice", "tri::n"], [0, 1, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri::n"], [0, 1, 1, "_CPPv43trii5Dtype14StreamOrDevice", "tri::s"], [0, 1, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri::s"], [0, 1, 1, "_CPPv43trii5Dtype14StreamOrDevice", "tri::type"], [0, 1, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri::type"], [0, 0, 1, "_CPPv44tril5arrayi14StreamOrDevice", "tril"], [0, 1, 1, "_CPPv44tril5arrayi14StreamOrDevice", "tril::k"], [0, 1, 1, "_CPPv44tril5arrayi14StreamOrDevice", "tril::s"], [0, 1, 1, "_CPPv44tril5arrayi14StreamOrDevice", "tril::x"], [0, 0, 1, "_CPPv44triu5arrayi14StreamOrDevice", "triu"], [0, 1, 1, "_CPPv44triu5arrayi14StreamOrDevice", "triu::k"], [0, 1, 1, "_CPPv44triu5arrayi14StreamOrDevice", "triu::s"], [0, 1, 1, "_CPPv44triu5arrayi14StreamOrDevice", "triu::x"], [0, 0, 1, "_CPPv43varRK5array14StreamOrDevice", "var"], [0, 0, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var"], [0, 0, 1, "_CPPv43varRK5arraybi14StreamOrDevice", "var"], [0, 0, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var"], [0, 1, 1, "_CPPv43varRK5array14StreamOrDevice", "var::a"], [0, 1, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var::a"], [0, 1, 1, "_CPPv43varRK5arraybi14StreamOrDevice", "var::a"], [0, 1, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var::a"], [0, 1, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var::axes"], [0, 1, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var::axis"], [0, 1, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var::ddof"], [0, 1, 1, "_CPPv43varRK5arraybi14StreamOrDevice", "var::ddof"], [0, 1, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var::ddof"], [0, 1, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var::keepdims"], [0, 1, 1, "_CPPv43varRK5arraybi14StreamOrDevice", "var::keepdims"], [0, 1, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var::keepdims"], [0, 1, 1, "_CPPv43varRK5array14StreamOrDevice", "var::s"], [0, 1, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var::s"], [0, 1, 1, "_CPPv43varRK5arraybi14StreamOrDevice", "var::s"], [0, 1, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var::s"], [0, 0, 1, "_CPPv44viewRK5arrayRK5Dtype14StreamOrDevice", "view"], [0, 1, 1, "_CPPv44viewRK5arrayRK5Dtype14StreamOrDevice", "view::a"], [0, 1, 1, "_CPPv44viewRK5arrayRK5Dtype14StreamOrDevice", "view::dtype"], [0, 1, 1, "_CPPv44viewRK5arrayRK5Dtype14StreamOrDevice", "view::s"], [0, 0, 1, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", "where"], [0, 1, 1, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", "where::condition"], [0, 1, 1, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", "where::s"], [0, 1, 1, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", "where::x"], [0, 1, 1, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", "where::y"], [0, 0, 1, "_CPPv45zerosRKNSt6vectorIiEE14StreamOrDevice", "zeros"], [0, 0, 1, "_CPPv45zerosRKNSt6vectorIiEE5Dtype14StreamOrDevice", "zeros"], [0, 1, 1, "_CPPv45zerosRKNSt6vectorIiEE5Dtype14StreamOrDevice", "zeros::dtype"], [0, 1, 1, "_CPPv45zerosRKNSt6vectorIiEE14StreamOrDevice", "zeros::s"], [0, 1, 1, "_CPPv45zerosRKNSt6vectorIiEE5Dtype14StreamOrDevice", "zeros::s"], [0, 1, 1, "_CPPv45zerosRKNSt6vectorIiEE14StreamOrDevice", "zeros::shape"], [0, 1, 1, "_CPPv45zerosRKNSt6vectorIiEE5Dtype14StreamOrDevice", "zeros::shape"], [0, 0, 1, "_CPPv410zeros_likeRK5array14StreamOrDevice", "zeros_like"], [0, 1, 1, "_CPPv410zeros_likeRK5array14StreamOrDevice", "zeros_like::a"], [0, 1, 1, "_CPPv410zeros_likeRK5array14StreamOrDevice", "zeros_like::s"]], "mlx.core": [[9, 3, 1, "", "Device"], [10, 3, 1, "", "Dtype"], [11, 3, 1, "", "DtypeCategory"], [315, 3, 1, "", "Stream"], [12, 5, 1, "", "abs"], [13, 5, 1, "", "add"], [14, 5, 1, "", "addmm"], [15, 5, 1, "", "all"], [16, 5, 1, "", "allclose"], [17, 5, 1, "", "any"], [18, 5, 1, "", "arange"], [19, 5, 1, "", "arccos"], [20, 5, 1, "", "arccosh"], [21, 5, 1, "", "arcsin"], [22, 5, 1, "", "arcsinh"], [23, 5, 1, "", "arctan"], [24, 5, 1, "", "arctan2"], [25, 5, 1, "", "arctanh"], [26, 5, 1, "", "argmax"], [27, 5, 1, "", "argmin"], [28, 5, 1, "", "argpartition"], [29, 5, 1, "", "argsort"], [30, 3, 1, "", "array"], [82, 5, 1, "", "array_equal"], [83, 5, 1, "", "as_strided"], [84, 5, 1, "", "atleast_1d"], [85, 5, 1, "", "atleast_2d"], [86, 5, 1, "", "atleast_3d"], [87, 5, 1, "", "bitwise_and"], [88, 5, 1, "", "bitwise_or"], [89, 5, 1, "", "bitwise_xor"], [90, 5, 1, "", "block_masked_mm"], [91, 5, 1, "", "broadcast_to"], [92, 5, 1, "", "ceil"], [93, 5, 1, "", "clip"], [94, 5, 1, "", "compile"], [95, 5, 1, "", "concatenate"], [96, 5, 1, "", "conj"], [97, 5, 1, "", "conjugate"], [98, 5, 1, "", "conv1d"], [99, 5, 1, "", "conv2d"], [100, 5, 1, "", "conv3d"], [101, 5, 1, "", "conv_general"], [102, 5, 1, "", "conv_transpose1d"], [103, 5, 1, "", "conv_transpose2d"], [104, 5, 1, "", "conv_transpose3d"], [105, 5, 1, "", "convolve"], [106, 5, 1, "", "cos"], [107, 5, 1, "", "cosh"], [108, 5, 1, "", "cummax"], [109, 5, 1, "", "cummin"], [110, 5, 1, "", "cumprod"], [111, 5, 1, "", "cumsum"], [112, 3, 1, "", "custom_function"], [113, 5, 1, "", "default_device"], [114, 5, 1, "", "default_stream"], [115, 5, 1, "", "degrees"], [116, 5, 1, "", "dequantize"], [117, 5, 1, "", "diag"], [118, 5, 1, "", "diagonal"], [119, 5, 1, "", "disable_compile"], [128, 5, 1, "", "divide"], [129, 5, 1, "", "divmod"], [130, 5, 1, "", "einsum"], [131, 5, 1, "", "einsum_path"], [132, 5, 1, "", "enable_compile"], [133, 5, 1, "", "equal"], [134, 5, 1, "", "erf"], [135, 5, 1, "", "erfinv"], [136, 5, 1, "", "eval"], [137, 5, 1, "", "exp"], [138, 5, 1, "", "expand_dims"], [139, 5, 1, "", "expm1"], [140, 5, 1, "", "eye"], [159, 5, 1, "", "flatten"], [160, 5, 1, "", "floor"], [161, 5, 1, "", "floor_divide"], [162, 5, 1, "", "full"], [163, 5, 1, "", "gather_mm"], [164, 5, 1, "", "gather_qmm"], [165, 5, 1, "", "grad"], [166, 5, 1, "", "greater"], [167, 5, 1, "", "greater_equal"], [168, 5, 1, "", "hadamard_transform"], [169, 5, 1, "", "identity"], [170, 5, 1, "", "imag"], [171, 5, 1, "", "inner"], [172, 5, 1, "", "isclose"], [173, 5, 1, "", "isfinite"], [174, 5, 1, "", "isinf"], [175, 5, 1, "", "isnan"], [176, 5, 1, "", "isneginf"], [177, 5, 1, "", "isposinf"], [178, 5, 1, "", "issubdtype"], [179, 5, 1, "", "jvp"], [180, 5, 1, "", "left_shift"], [181, 5, 1, "", "less"], [182, 5, 1, "", "less_equal"], [193, 5, 1, "", "linspace"], [194, 5, 1, "", "load"], [195, 5, 1, "", "log"], [196, 5, 1, "", "log10"], [197, 5, 1, "", "log1p"], [198, 5, 1, "", "log2"], [199, 5, 1, "", "logaddexp"], [200, 5, 1, "", "logical_and"], [201, 5, 1, "", "logical_not"], [202, 5, 1, "", "logical_or"], [203, 5, 1, "", "logsumexp"], [204, 5, 1, "", "matmul"], [205, 5, 1, "", "max"], [206, 5, 1, "", "maximum"], [207, 5, 1, "", "mean"], [208, 5, 1, "", "meshgrid"], [221, 5, 1, "", "min"], [222, 5, 1, "", "minimum"], [223, 5, 1, "", "moveaxis"], [224, 5, 1, "", "multiply"], [225, 5, 1, "", "nan_to_num"], [226, 5, 1, "", "negative"], [227, 5, 1, "", "new_stream"], [228, 5, 1, "", "not_equal"], [229, 5, 1, "", "ones"], [230, 5, 1, "", "ones_like"], [231, 5, 1, "", "outer"], [232, 5, 1, "", "pad"], [233, 5, 1, "", "partition"], [234, 5, 1, "", "power"], [235, 5, 1, "", "prod"], [236, 5, 1, "", "put_along_axis"], [237, 5, 1, "", "quantize"], [238, 5, 1, "", "quantized_matmul"], [239, 5, 1, "", "radians"], [253, 5, 1, "", "real"], [254, 5, 1, "", "reciprocal"], [255, 5, 1, "", "remainder"], [256, 5, 1, "", "repeat"], [257, 5, 1, "", "reshape"], [258, 5, 1, "", "right_shift"], [259, 5, 1, "", "roll"], [260, 5, 1, "", "round"], [261, 5, 1, "", "rsqrt"], [262, 5, 1, "", "save"], [263, 5, 1, "", "save_gguf"], [264, 5, 1, "", "save_safetensors"], [265, 5, 1, "", "savez"], [266, 5, 1, "", "savez_compressed"], [267, 5, 1, "", "set_default_device"], [268, 5, 1, "", "set_default_stream"], [269, 5, 1, "", "sigmoid"], [270, 5, 1, "", "sign"], [271, 5, 1, "", "sin"], [272, 5, 1, "", "sinh"], [273, 5, 1, "", "softmax"], [274, 5, 1, "", "sort"], [275, 5, 1, "", "split"], [276, 5, 1, "", "sqrt"], [277, 5, 1, "", "square"], [278, 5, 1, "", "squeeze"], [279, 5, 1, "", "stack"], [280, 5, 1, "", "std"], [281, 5, 1, "", "stop_gradient"], [282, 5, 1, "", "stream"], [283, 5, 1, "", "subtract"], [284, 5, 1, "", "sum"], [285, 5, 1, "", "swapaxes"], [286, 5, 1, "", "synchronize"], [287, 5, 1, "", "take"], [288, 5, 1, "", "take_along_axis"], [289, 5, 1, "", "tan"], [290, 5, 1, "", "tanh"], [291, 5, 1, "", "tensordot"], [292, 5, 1, "", "tile"], [293, 5, 1, "", "topk"], [294, 5, 1, "", "trace"], [295, 5, 1, "", "transpose"], [296, 5, 1, "", "tri"], [297, 5, 1, "", "tril"], [298, 5, 1, "", "triu"], [299, 5, 1, "", "value_and_grad"], [300, 5, 1, "", "var"], [301, 5, 1, "", "view"], [302, 5, 1, "", "vjp"], [303, 5, 1, "", "vmap"], [304, 5, 1, "", "where"], [305, 5, 1, "", "zeros"], [306, 5, 1, "", "zeros_like"]], "mlx.core.Device": [[9, 4, 1, "", "__init__"]], "mlx.core.Dtype": [[10, 4, 1, "", "__init__"]], "mlx.core.DtypeCategory": [[11, 4, 1, "", "__init__"]], "mlx.core.Stream": [[315, 4, 1, "", "__init__"]], "mlx.core.array": [[31, 6, 1, "", "T"], [30, 4, 1, "", "__init__"], [32, 4, 1, "", "abs"], [33, 4, 1, "", "all"], [34, 4, 1, "", "any"], [35, 4, 1, "", "argmax"], [36, 4, 1, "", "argmin"], [37, 4, 1, "", "astype"], [38, 6, 1, "", "at"], [39, 4, 1, "", "conj"], [40, 4, 1, "", "cos"], [41, 4, 1, "", "cummax"], [42, 4, 1, "", "cummin"], [43, 4, 1, "", "cumprod"], [44, 4, 1, "", "cumsum"], [45, 4, 1, "", "diag"], [46, 4, 1, "", "diagonal"], [47, 6, 1, "", "dtype"], [48, 4, 1, "", "exp"], [49, 4, 1, "", "flatten"], [50, 4, 1, "", "item"], [51, 6, 1, "", "itemsize"], [52, 4, 1, "", "log"], [53, 4, 1, "", "log10"], [54, 4, 1, "", "log1p"], [55, 4, 1, "", "log2"], [56, 4, 1, "", "logsumexp"], [57, 4, 1, "", "max"], [58, 4, 1, "", "mean"], [59, 4, 1, "", "min"], [60, 4, 1, "", "moveaxis"], [61, 6, 1, "", "nbytes"], [62, 6, 1, "", "ndim"], [63, 4, 1, "", "prod"], [64, 4, 1, "", "reciprocal"], [65, 4, 1, "", "reshape"], [66, 4, 1, "", "round"], [67, 4, 1, "", "rsqrt"], [68, 6, 1, "", "shape"], [69, 4, 1, "", "sin"], [70, 6, 1, "", "size"], [71, 4, 1, "", "split"], [72, 4, 1, "", "sqrt"], [73, 4, 1, "", "square"], [74, 4, 1, "", "squeeze"], [75, 4, 1, "", "std"], [76, 4, 1, "", "sum"], [77, 4, 1, "", "swapaxes"], [78, 4, 1, "", "tolist"], [79, 4, 1, "", "transpose"], [80, 4, 1, "", "var"], [81, 4, 1, "", "view"]], "mlx.core.custom_function": [[112, 4, 1, "", "__init__"]], "mlx.core.distributed": [[120, 3, 1, "", "Group"], [121, 5, 1, "", "all_gather"], [122, 5, 1, "", "all_sum"], [123, 5, 1, "", "init"], [124, 5, 1, "", "is_available"], [125, 5, 1, "", "recv"], [126, 5, 1, "", "recv_like"], [127, 5, 1, "", "send"]], "mlx.core.distributed.Group": [[120, 4, 1, "", "__init__"]], "mlx.core.fast": [[141, 5, 1, "", "affine_quantize"], [142, 5, 1, "", "layer_norm"], [143, 5, 1, "", "metal_kernel"], [144, 5, 1, "", "rms_norm"], [145, 5, 1, "", "rope"], [146, 5, 1, "", "scaled_dot_product_attention"]], "mlx.core.fft": [[147, 5, 1, "", "fft"], [148, 5, 1, "", "fft2"], [149, 5, 1, "", "fftn"], [150, 5, 1, "", "ifft"], [151, 5, 1, "", "ifft2"], [152, 5, 1, "", "ifftn"], [153, 5, 1, "", "irfft"], [154, 5, 1, "", "irfft2"], [155, 5, 1, "", "irfftn"], [156, 5, 1, "", "rfft"], [157, 5, 1, "", "rfft2"], [158, 5, 1, "", "rfftn"]], "mlx.core.linalg": [[183, 5, 1, "", "cholesky"], [184, 5, 1, "", "cholesky_inv"], [185, 5, 1, "", "cross"], [186, 5, 1, "", "eigh"], [187, 5, 1, "", "eigvalsh"], [188, 5, 1, "", "inv"], [189, 5, 1, "", "norm"], [190, 5, 1, "", "qr"], [191, 5, 1, "", "svd"], [192, 5, 1, "", "tri_inv"]], "mlx.core.metal": [[209, 5, 1, "", "clear_cache"], [210, 5, 1, "", "device_info"], [211, 5, 1, "", "get_active_memory"], [212, 5, 1, "", "get_cache_memory"], [213, 5, 1, "", "get_peak_memory"], [214, 5, 1, "", "is_available"], [215, 5, 1, "", "reset_peak_memory"], [216, 5, 1, "", "set_cache_limit"], [217, 5, 1, "", "set_memory_limit"], [218, 5, 1, "", "set_wired_limit"], [219, 5, 1, "", "start_capture"], [220, 5, 1, "", "stop_capture"]], "mlx.core.random": [[240, 5, 1, "", "bernoulli"], [241, 5, 1, "", "categorical"], [242, 5, 1, "", "gumbel"], [243, 5, 1, "", "key"], [244, 5, 1, "", "laplace"], [245, 5, 1, "", "multivariate_normal"], [246, 5, 1, "", "normal"], [247, 5, 1, "", "permutation"], [248, 5, 1, "", "randint"], [249, 5, 1, "", "seed"], [250, 5, 1, "", "split"], [251, 5, 1, "", "truncated_normal"], [252, 5, 1, "", "uniform"]], "mlx.nn": [[325, 3, 1, "", "ALiBi"], [326, 3, 1, "", "AvgPool1d"], [327, 3, 1, "", "AvgPool2d"], [328, 3, 1, "", "BatchNorm"], [329, 3, 1, "", "CELU"], [330, 3, 1, "", "Conv1d"], [331, 3, 1, "", "Conv2d"], [332, 3, 1, "", "Conv3d"], [333, 3, 1, "", "ConvTranspose1d"], [334, 3, 1, "", "ConvTranspose2d"], [335, 3, 1, "", "ConvTranspose3d"], [336, 3, 1, "", "Dropout"], [337, 3, 1, "", "Dropout2d"], [338, 3, 1, "", "Dropout3d"], [339, 3, 1, "", "ELU"], [340, 3, 1, "", "Embedding"], [341, 3, 1, "", "GELU"], [342, 3, 1, "", "GLU"], [343, 3, 1, "", "GRU"], [344, 3, 1, "", "GroupNorm"], [345, 3, 1, "", "HardShrink"], [346, 3, 1, "", "HardTanh"], [347, 3, 1, "", "Hardswish"], [348, 3, 1, "", "InstanceNorm"], [349, 3, 1, "", "LSTM"], [350, 3, 1, "", "LayerNorm"], [351, 3, 1, "", "LeakyReLU"], [352, 3, 1, "", "Linear"], [353, 3, 1, "", "LogSigmoid"], [354, 3, 1, "", "LogSoftmax"], [355, 3, 1, "", "MaxPool1d"], [356, 3, 1, "", "MaxPool2d"], [357, 3, 1, "", "Mish"], [452, 3, 1, "", "Module"], [378, 3, 1, "", "MultiHeadAttention"], [379, 3, 1, "", "PReLU"], [380, 3, 1, "", "QuantizedEmbedding"], [381, 3, 1, "", "QuantizedLinear"], [382, 3, 1, "", "RMSNorm"], [383, 3, 1, "", "RNN"], [384, 3, 1, "", "ReLU"], [385, 3, 1, "", "ReLU6"], [386, 3, 1, "", "RoPE"], [387, 3, 1, "", "SELU"], [388, 3, 1, "", "Sequential"], [389, 3, 1, "", "SiLU"], [390, 3, 1, "", "Sigmoid"], [391, 3, 1, "", "SinusoidalPositionalEncoding"], [392, 3, 1, "", "Softmax"], [393, 3, 1, "", "Softmin"], [394, 3, 1, "", "Softplus"], [395, 3, 1, "", "Softshrink"], [396, 3, 1, "", "Softsign"], [397, 3, 1, "", "Step"], [398, 3, 1, "", "Tanh"], [399, 3, 1, "", "Transformer"], [400, 3, 1, "", "Upsample"], [409, 3, 1, "", "celu"], [410, 3, 1, "", "elu"], [411, 3, 1, "", "gelu"], [412, 3, 1, "", "gelu_approx"], [413, 3, 1, "", "gelu_fast_approx"], [414, 3, 1, "", "glu"], [415, 3, 1, "", "hard_shrink"], [416, 3, 1, "", "hard_tanh"], [417, 3, 1, "", "hardswish"], [418, 3, 1, "", "leaky_relu"], [419, 3, 1, "", "log_sigmoid"], [420, 3, 1, "", "log_softmax"], [435, 3, 1, "", "mish"], [436, 3, 1, "", "prelu"], [307, 5, 1, "", "quantize"], [437, 3, 1, "", "relu"], [438, 3, 1, "", "relu6"], [439, 3, 1, "", "selu"], [440, 3, 1, "", "sigmoid"], [441, 3, 1, "", "silu"], [442, 3, 1, "", "softmax"], [443, 3, 1, "", "softmin"], [444, 3, 1, "", "softplus"], [445, 3, 1, "", "softshrink"], [446, 3, 1, "", "step"], [447, 3, 1, "", "tanh"], [308, 5, 1, "", "value_and_grad"]], "mlx.nn.Module": [[358, 4, 1, "", "apply"], [359, 4, 1, "", "apply_to_modules"], [360, 4, 1, "", "children"], [361, 4, 1, "", "eval"], [362, 4, 1, "", "filter_and_map"], [363, 4, 1, "", "freeze"], [364, 4, 1, "", "leaf_modules"], [365, 4, 1, "", "load_weights"], [366, 4, 1, "", "modules"], [367, 4, 1, "", "named_modules"], [368, 4, 1, "", "parameters"], [369, 4, 1, "", "save_weights"], [370, 4, 1, "", "set_dtype"], [371, 6, 1, "", "state"], [372, 4, 1, "", "train"], [373, 4, 1, "", "trainable_parameters"], [374, 6, 1, "", "training"], [375, 4, 1, "", "unfreeze"], [376, 4, 1, "", "update"], [377, 4, 1, "", "update_modules"]], "mlx.nn.init": [[401, 5, 1, "", "constant"], [402, 5, 1, "", "glorot_normal"], [403, 5, 1, "", "glorot_uniform"], [404, 5, 1, "", "he_normal"], [405, 5, 1, "", "he_uniform"], [406, 5, 1, "", "identity"], [407, 5, 1, "", "normal"], [408, 5, 1, "", "uniform"]], "mlx.nn.losses": [[421, 3, 1, "", "binary_cross_entropy"], [422, 3, 1, "", "cosine_similarity_loss"], [423, 3, 1, "", "cross_entropy"], [424, 3, 1, "", "gaussian_nll_loss"], [425, 3, 1, "", "hinge_loss"], [426, 3, 1, "", "huber_loss"], [427, 3, 1, "", "kl_div_loss"], [428, 3, 1, "", "l1_loss"], [429, 3, 1, "", "log_cosh_loss"], [430, 3, 1, "", "margin_ranking_loss"], [431, 3, 1, "", "mse_loss"], [432, 3, 1, "", "nll_loss"], [433, 3, 1, "", "smooth_l1_loss"], [434, 3, 1, "", "triplet_loss"]], "mlx.optimizers": [[455, 3, 1, "", "AdaDelta"], [456, 3, 1, "", "Adafactor"], [457, 3, 1, "", "Adagrad"], [458, 3, 1, "", "Adam"], [459, 3, 1, "", "AdamW"], [460, 3, 1, "", "Adamax"], [461, 3, 1, "", "Lion"], [474, 3, 1, "", "Optimizer"], [466, 3, 1, "", "RMSprop"], [467, 3, 1, "", "SGD"], [309, 5, 1, "", "clip_grad_norm"], [468, 5, 1, "", "cosine_decay"], [469, 5, 1, "", "exponential_decay"], [470, 5, 1, "", "join_schedules"], [471, 5, 1, "", "linear_schedule"], [472, 5, 1, "", "step_decay"]], "mlx.optimizers.Optimizer": [[462, 4, 1, "", "apply_gradients"], [463, 4, 1, "", "init"], [464, 6, 1, "", "state"], [465, 4, 1, "", "update"]], "mlx.utils": [[310, 5, 1, "", "tree_flatten"], [311, 5, 1, "", "tree_map"], [312, 5, 1, "", "tree_map_with_path"], [313, 5, 1, "", "tree_reduce"], [314, 5, 1, "", "tree_unflatten"]]}, "objnames": {"0": ["cpp", "function", "C++ function"], "1": ["cpp", "functionParam", "C++ function parameter"], "2": ["cpp", "templateParam", "C++ template parameter"], "3": ["py", "class", "Python class"], "4": ["py", "method", "Python method"], "5": ["py", "function", "Python function"], "6": ["py", "property", "Python property"]}, "objtypes": {"0": "cpp:function", "1": "cpp:functionParam", "2": "cpp:templateParam", "3": "py:class", "4": "py:method", "5": "py:function", "6": "py:property"}, "terms": {"": [0, 1, 2, 4, 5, 6, 47, 51, 62, 94, 114, 116, 141, 148, 149, 151, 152, 154, 155, 157, 158, 165, 184, 189, 191, 194, 207, 231, 237, 241, 260, 263, 264, 280, 282, 299, 300, 301, 303, 308, 324, 327, 343, 349, 356, 362, 363, 365, 369, 370, 371, 375, 383, 454, 463, 464, 476, 479, 481, 483, 484, 485, 486, 487], "0": [0, 1, 2, 4, 5, 6, 8, 9, 14, 18, 38, 45, 46, 49, 66, 71, 75, 80, 83, 95, 98, 99, 100, 101, 102, 103, 104, 117, 118, 140, 143, 146, 159, 163, 165, 186, 188, 189, 190, 192, 209, 216, 218, 225, 232, 240, 244, 246, 247, 252, 256, 260, 275, 279, 280, 294, 296, 297, 298, 299, 300, 303, 309, 310, 312, 313, 324, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 344, 345, 348, 350, 351, 355, 356, 379, 384, 386, 391, 395, 397, 399, 401, 402, 403, 404, 405, 406, 407, 408, 409, 410, 412, 413, 415, 416, 417, 418, 421, 423, 425, 426, 430, 433, 434, 436, 437, 438, 439, 445, 446, 449, 452, 455, 456, 458, 459, 460, 461, 463, 466, 467, 468, 469, 470, 471, 472, 476, 479, 480, 481, 482, 483, 484, 485, 486], "00005": 4, "0001": 391, "0005": 412, "001": 456, "00364": 4, "01": [4, 351, 418, 459], "0137595": 404, "015": 413, "0184009": 405, "02264": 403, "025": 481, "02765": 404, "0300242": 405, "044715": [341, 412], "0485873": 423, "05": [16, 172, 328, 344, 348, 350, 382], "0507": 439, "05202": 5, "06": [424, 434, 455], "0638": 430, "06450": 350, "0645099": 407, "06561": 469, "06675": 461, "07467": 382, "08": [16, 172, 422, 457, 458, 459, 460, 466], "08022": 348, "081": 472, "08415": 413, "08494": 344, "08619": 405, "08681": [357, 435], "09864": 5, "0999938": 470, "0999961": 468, "0f": 0, "1": [0, 1, 2, 3, 5, 6, 14, 18, 28, 29, 38, 46, 49, 98, 99, 100, 101, 102, 103, 104, 117, 118, 139, 143, 146, 147, 148, 150, 151, 153, 154, 155, 156, 157, 158, 159, 168, 171, 178, 184, 185, 186, 187, 189, 190, 204, 208, 217, 231, 233, 237, 241, 244, 245, 246, 252, 269, 274, 287, 293, 294, 299, 309, 312, 313, 317, 324, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 342, 343, 344, 348, 349, 350, 352, 355, 356, 379, 382, 383, 386, 390, 391, 397, 400, 402, 403, 404, 405, 406, 407, 408, 409, 410, 412, 413, 414, 416, 419, 420, 421, 422, 423, 424, 425, 426, 427, 429, 430, 432, 433, 434, 439, 440, 442, 443, 444, 446, 449, 452, 454, 455, 456, 457, 458, 459, 460, 461, 463, 466, 467, 468, 469, 470, 471, 472, 479, 480, 481, 482, 484, 485, 486, 487], "10": [0, 3, 5, 6, 196, 260, 265, 311, 324, 365, 449, 470, 472, 479, 480, 482], "100": [2, 4, 5, 421, 471, 479, 481, 483, 487], "1000": [468, 479], "10000": 386, "101": 471, "1024": [1, 5], "105361": 421, "109": 2, "10_000": 4, "10x": 461, "11": 189, "114": 2, "12": [5, 168, 470], "1212": 455, "12451": 403, "128": [265, 324], "13": 8, "14": 8, "15": [1, 8, 189, 218, 313, 479], "150594": 402, "15268": 404, "16": [1, 143, 317, 326, 348, 355, 358, 452], "1606": 413, "1607": [348, 350], "16384": 168, "16506": 405, "17": 8, "177208": 404, "1803": 344, "1908": [357, 435], "1910": 382, "191107": 402, "1985": 189, "1_000": 4, "1d": [0, 98, 102, 105, 263, 288], "1e": [0, 4, 6, 16, 172, 328, 344, 348, 350, 351, 382, 422, 424, 434, 454, 455, 456, 457, 458, 459, 460, 463, 466, 468, 469, 470, 471, 472], "1e3": 479, "1st": 237, "2": [0, 1, 2, 4, 5, 6, 38, 99, 103, 117, 118, 134, 148, 151, 153, 154, 155, 156, 157, 158, 159, 168, 178, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 198, 204, 237, 245, 250, 291, 294, 296, 297, 298, 309, 313, 317, 324, 326, 327, 331, 334, 341, 351, 355, 356, 382, 391, 400, 401, 402, 403, 404, 405, 406, 407, 408, 412, 423, 424, 426, 433, 434, 449, 452, 454, 455, 457, 458, 459, 463, 466, 479, 480, 481, 482, 483, 484, 485, 486, 487], "20": [168, 189], "200": [5, 470], "2002": 5, "2011": 457, "2012": [455, 466], "2015": [337, 458, 460], "2019": [5, 459], "2020": 5, "2021": 5, "20397": 421, "20_000": 5, "21": [5, 472], "2104": 5, "223144": 421, "223404": 403, "225": 189, "225763": 430, "2302": 461, "23607": [189, 190], "24": 8, "24264": 189, "247": 5, "25": [379, 400], "25211": 404, "256": [1, 2, 6, 143], "256995": 430, "28": 168, "2d": [0, 99, 103, 118, 237, 328, 337], "2nd": 237, "2x": 484, "3": [0, 1, 2, 5, 8, 100, 104, 159, 178, 185, 186, 187, 189, 190, 309, 313, 332, 335, 341, 400, 403, 405, 412, 417, 456, 461, 476, 479, 482, 484, 485], "30": 456, "3118": 484, "32": [1, 5, 6, 90, 237, 238, 317, 327, 356, 382, 479], "330": 5, "33333": 400, "348587": 423, "363207": 402, "36788": 479, "379159": 403, "380709": 407, "39": 5, "390": 481, "3d": [0, 2, 100, 104, 328, 338, 400], "3f": [2, 6, 479], "3x": 2, "4": [0, 1, 2, 5, 116, 141, 143, 159, 164, 189, 237, 238, 265, 307, 313, 317, 326, 327, 328, 348, 355, 356, 380, 381, 399, 400, 402, 403, 404, 421, 479, 480, 482, 485, 487], "4096": [479, 481, 487], "40x": 1, "41421": 189, "417497": 408, "42": 314, "437": 5, "44": 5, "447214": 190, "458835": 404, "475": 5, "48095": 402, "4d": [1, 400], "4m": 1, "5": [0, 1, 2, 4, 5, 8, 189, 217, 240, 313, 326, 328, 336, 337, 338, 341, 345, 348, 355, 395, 400, 401, 404, 405, 412, 415, 433, 445, 449, 454, 466, 468, 469, 479, 481, 482], "50": [0, 193], "500": [5, 487], "5000": 2, "510826": 421, "512": [2, 3, 5, 399, 487], "534422": 407, "539245": 421, "53947": 402, "55": 1, "5701": 455, "573409": 430, "57771": 190, "579": 5, "5f": 4, "6": [1, 2, 5, 189, 265, 385, 399, 403, 412, 413, 417, 424, 434, 438, 466, 479, 482, 485], "61278": 402, "617261": 408, "628": 5, "633": 5, "64": [0, 1, 90, 116, 141, 164, 237, 238, 307, 317, 380, 381], "64331": 405, "666329": 405, "66667": 400, "67326": 439, "676": 1, "690": 5, "6967": 404, "7": [2, 5, 189, 237, 482], "702": [341, 413], "707107": 186, "71828": 479, "74166": 189, "74597": 189, "75": 400, "75596": 430, "75787": 404, "765166": 430, "773433": 430, "776856": 403, "793615": 405, "79854": 405, "7b": 5, "7m": 1, "8": [0, 1, 2, 5, 8, 189, 237, 317, 327, 348, 356, 399, 422, 455, 456, 457, 458, 459, 460, 466, 479, 482, 485, 487], "8192": [5, 168], "84804": 189, "863726": 408, "883935": 408, "890597": 403, "894427": 190, "89613": 402, "8gb": 5, "8x": 1, "9": [8, 189, 423, 455, 458, 459, 460, 461, 463, 469, 472, 484], "90041": 403, "912766": 403, "916291": 421, "95": 6, "982273": 407, "99": [461, 466], "995016": 402, "999": [458, 459, 460], "A": [0, 2, 5, 7, 8, 9, 68, 82, 94, 142, 143, 144, 146, 165, 178, 179, 184, 186, 187, 189, 190, 191, 194, 203, 204, 205, 210, 221, 237, 240, 241, 242, 244, 245, 246, 247, 248, 251, 252, 275, 279, 282, 299, 302, 303, 307, 308, 309, 310, 311, 312, 313, 314, 315, 324, 328, 337, 343, 344, 348, 350, 362, 366, 367, 370, 376, 377, 382, 388, 391, 399, 402, 403, 405, 413, 434, 435, 452, 454, 458, 460, 462, 463, 465, 470, 479, 480, 481, 483, 484], "AS": 163, "And": [5, 400], "As": [6, 38, 287, 324], "At": 93, "But": 487, "By": [5, 307, 370, 421, 481, 484], "For": [0, 1, 2, 5, 8, 38, 146, 163, 178, 189, 237, 314, 324, 328, 337, 341, 358, 363, 372, 375, 381, 386, 391, 400, 402, 403, 404, 405, 421, 449, 454, 476, 479, 480, 481, 482, 483, 484, 485, 486, 487], "If": [0, 1, 2, 5, 8, 15, 16, 17, 18, 26, 27, 28, 29, 78, 82, 83, 93, 95, 105, 108, 109, 110, 111, 117, 118, 121, 122, 123, 125, 126, 127, 136, 142, 145, 156, 157, 158, 161, 162, 165, 172, 183, 184, 185, 189, 194, 203, 204, 205, 207, 208, 216, 217, 221, 225, 229, 232, 233, 235, 236, 241, 245, 247, 256, 259, 273, 274, 275, 280, 284, 286, 287, 288, 291, 293, 294, 299, 300, 303, 305, 307, 311, 313, 328, 330, 331, 332, 333, 334, 335, 344, 350, 352, 363, 365, 375, 381, 383, 386, 388, 391, 400, 421, 423, 434, 456, 479, 480, 481, 483, 486, 487, 488], "In": [0, 1, 2, 5, 6, 38, 204, 237, 311, 324, 337, 344, 452, 455, 457, 458, 460, 461, 462, 478, 479, 480, 481, 483, 486, 487], "It": [2, 5, 8, 126, 165, 268, 299, 309, 313, 324, 377, 381, 462, 474, 484, 486], "Its": 324, "No": [2, 5, 186, 187], "Not": [94, 228, 479], "ON": [3, 8], "Of": 481, "On": [1, 479, 481, 483], "One": [147, 150, 156, 232, 261, 479, 481], "THE": 8, "That": 5, "The": [0, 1, 2, 3, 5, 6, 7, 8, 12, 13, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 37, 47, 51, 61, 62, 68, 78, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 115, 116, 117, 118, 121, 122, 123, 125, 126, 127, 128, 129, 130, 131, 133, 134, 135, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 179, 180, 181, 182, 185, 186, 187, 189, 190, 191, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 212, 213, 216, 217, 218, 219, 221, 222, 223, 224, 226, 228, 229, 230, 231, 232, 233, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 263, 264, 269, 270, 271, 272, 273, 274, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 307, 308, 309, 310, 311, 312, 313, 314, 317, 319, 326, 327, 328, 330, 331, 332, 333, 334, 335, 336, 337, 338, 340, 342, 343, 344, 348, 349, 350, 352, 355, 356, 358, 359, 363, 365, 369, 370, 371, 372, 375, 376, 377, 378, 380, 381, 382, 383, 386, 388, 391, 397, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 414, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 446, 449, 452, 454, 455, 456, 457, 458, 459, 460, 461, 464, 466, 467, 468, 471, 474, 479, 480, 481, 482, 483, 484, 485, 486, 487, 488], "Then": [4, 8], "There": [1, 2, 324, 400, 479], "These": [1, 2, 94, 236, 288, 423, 487], "To": [0, 2, 3, 4, 5, 6, 8, 216, 324, 449, 454, 479, 480, 481, 485], "With": 2, "_": [1, 3, 4, 5, 312, 324, 468, 469, 470, 471, 472, 476, 479, 483, 487], "__call__": [1, 5, 6, 324, 452], "__init__": [2, 5, 6, 9, 10, 11, 30, 112, 120, 315, 324, 452], "__main__": [2, 5], "__name__": [2, 5], "_a": 2, "_ext": 2, "_f": 189, "_in": [402, 403], "_out": [402, 403], "_p": 434, "_size": [326, 327, 355, 356], "_val": 416, "a1": 163, "a2": 163, "a_": 189, "a_max": [0, 93], "a_min": [0, 93], "a_ndim": 1, "a_shap": 1, "a_strid": 1, "a_view": 484, "ab": [0, 16, 172, 189, 299, 344, 348, 350, 357, 382, 413, 435, 479], "abil": 480, "abl": [2, 237], "about": [1, 2, 5, 6, 131, 210, 483, 487], "abov": [1, 2, 5, 237, 297, 324, 400, 459, 480, 481, 482, 483, 487], "absolut": [0, 12, 16, 172, 412, 413, 433], "acc": 313, "acceler": [2, 328], "access": [0, 5, 50, 324, 452, 463, 480, 483, 487], "accord": [0, 242, 304, 307, 378, 402, 403, 404, 405], "accordingli": 2, "accross": 8, "accumul": [313, 382], "accuraci": 6, "accustom": 5, "achiev": [324, 480], "across": [1, 2, 344, 480], "act": [2, 429], "action": 324, "activ": [2, 8, 211, 337, 397, 399, 415, 435, 445, 446, 448, 479], "actual": [5, 18, 365, 452, 483], "ad": [0, 1, 2, 4, 8, 142, 348, 452, 455, 456, 457, 458, 459, 460, 466, 480, 483, 486], "adadelta": 454, "adafactor": 454, "adagrad": 454, "adam": [454, 460, 461, 470, 471], "adamax": 454, "adamw": [454, 461], "adapt": [455, 456, 457, 480], "add": [0, 1, 2, 3, 5, 14, 38, 138, 199, 232, 237, 330, 331, 332, 333, 334, 335, 481, 487], "add_argu": 5, "add_depend": 2, "add_librari": 2, "addit": [0, 2, 5, 8, 13, 14, 142, 144, 146, 194, 328, 344, 350, 378, 382, 452, 481], "addmm": 0, "address": 2, "adjac": 337, "advanc": [5, 479], "advantag": 487, "advis": 484, "affin": [328, 344, 348, 350, 352, 381], "after": [2, 5, 6, 28, 159, 161, 164, 209, 233, 237, 328, 344, 350, 358, 359, 363, 365, 372, 375, 376, 377, 378, 399, 433, 479, 487], "after_1": 232, "after_2": 232, "after_i": 232, "after_n": 232, "afternoon": 5, "again": [5, 8, 324, 479], "against": 0, "aggreg": 378, "ago": 5, "ai": 112, "ainv": [188, 192], "albeit": 487, "algebra": 7, "algorithm": [400, 461], "alia": [96, 97, 341], "alibi": 324, "align": [184, 237, 327, 343, 349, 356], "align_corn": 400, "all": [0, 1, 2, 3, 6, 8, 16, 28, 38, 84, 85, 86, 94, 99, 100, 101, 103, 104, 112, 121, 122, 123, 140, 149, 152, 155, 158, 163, 164, 191, 204, 232, 233, 259, 278, 307, 324, 358, 359, 363, 366, 367, 368, 373, 375, 378, 391, 399, 400, 449, 452, 474, 476, 479, 482, 483, 485, 488], "all_avg": 480, "all_reduce_grad": 480, "all_sum": 480, "allclos": [0, 1, 143], "alloc": [2, 212, 216, 217, 452], "allow": [0, 1, 2, 178, 309, 324, 377, 452, 474, 480, 482, 485], "almost": 5, "alon": [2, 484], "along": [0, 2, 26, 27, 94, 95, 108, 109, 110, 111, 121, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 163, 164, 168, 185, 189, 236, 247, 256, 259, 273, 275, 279, 287, 288, 291, 292, 293, 294, 301, 324, 342, 383, 414], "alpha": [0, 2, 14, 237, 329, 339, 409, 410, 434, 436, 439, 459, 466], "alpha_": 2, "alreadi": [2, 3, 5, 480], "also": [0, 1, 2, 5, 6, 7, 8, 11, 13, 87, 88, 89, 119, 128, 129, 133, 149, 152, 155, 158, 166, 167, 180, 181, 182, 199, 206, 222, 224, 228, 234, 237, 255, 258, 283, 307, 308, 319, 324, 362, 376, 378, 380, 381, 389, 411, 439, 441, 448, 454, 479, 480, 481, 482, 483, 484, 485, 488], "altern": 476, "alwai": [1, 83, 211, 310, 481], "am": 5, "among": 2, "amount": [5, 213, 326, 355], "amus": 5, "an": [0, 1, 2, 3, 5, 6, 8, 10, 15, 17, 30, 84, 85, 86, 91, 98, 99, 100, 101, 102, 103, 104, 120, 125, 126, 127, 136, 140, 142, 146, 159, 162, 169, 173, 183, 189, 194, 217, 218, 223, 229, 230, 232, 235, 236, 237, 238, 247, 256, 257, 259, 260, 275, 278, 285, 287, 288, 291, 292, 296, 303, 305, 306, 310, 311, 312, 313, 324, 326, 327, 336, 341, 344, 349, 350, 352, 355, 356, 358, 378, 379, 381, 383, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 412, 436, 449, 454, 455, 465, 469, 474, 476, 478, 479, 480, 481, 482, 483, 484, 485, 486, 487, 488], "anaconda": 480, "anchor": 434, "angl": [115, 239, 351], "angular": [145, 386], "ani": [0, 1, 2, 5, 7, 18, 94, 310, 311, 312, 313, 314, 324, 341, 358, 359, 362, 371, 381, 399, 400, 449, 471, 478, 479, 481, 483, 485, 486, 487], "anonym": 479, "anoth": [0, 93, 178, 204, 283, 304, 317, 324, 358, 479, 481, 482, 487], "anwywher": 8, "anyhow": 5, "anymor": 5, "anyth": [5, 299, 483], "anytim": 483, "api": [1, 2, 341, 480, 481], "app": 8, "append": [5, 204, 479, 483], "appl": [2, 5, 7, 8, 487], "appli": [0, 38, 145, 146, 163, 191, 311, 312, 313, 324, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 337, 338, 339, 341, 342, 344, 345, 346, 347, 348, 350, 351, 352, 353, 354, 355, 356, 357, 359, 372, 379, 381, 382, 383, 384, 385, 387, 389, 390, 392, 393, 394, 395, 396, 397, 398, 400, 409, 410, 411, 414, 415, 416, 417, 418, 419, 420, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445, 446, 447, 449, 462, 465, 471, 474, 479, 480], "applic": [3, 8], "apply_fn": 359, "apply_gradi": 454, "apply_to_modul": [324, 363], "approach": [429, 481], "appropri": [2, 479], "approx": 341, "approxim": [16, 341, 411, 412, 413], "ar": [0, 1, 2, 4, 5, 6, 7, 8, 16, 18, 82, 90, 91, 93, 94, 101, 105, 112, 118, 125, 126, 136, 140, 143, 148, 149, 151, 152, 154, 155, 157, 158, 159, 164, 165, 172, 173, 174, 175, 176, 177, 178, 179, 186, 187, 189, 190, 194, 204, 217, 231, 232, 233, 237, 238, 240, 241, 242, 247, 248, 251, 252, 259, 265, 266, 278, 279, 287, 299, 302, 303, 307, 310, 311, 317, 328, 330, 331, 332, 333, 334, 335, 336, 337, 338, 344, 348, 350, 352, 365, 378, 381, 400, 421, 423, 424, 448, 452, 454, 461, 463, 478, 479, 480, 481, 482, 483, 484, 485, 486, 487], "arang": [0, 1, 189, 247, 317, 400, 482, 484], "arbitrari": [310, 452], "arbitrarili": [1, 94, 324, 478, 481, 485], "arc": 0, "arcco": 0, "arccosh": 0, "architectur": [5, 8, 210, 324, 377, 487], "archiv": 486, "arcsin": 0, "arcsinh": 0, "arctan": 0, "arctan2": 0, "arctanh": 0, "arg": [2, 5, 10, 18, 120, 136, 265, 266], "arg1": 178, "arg2": 178, "argmax": [0, 6], "argmin": 0, "argnam": [165, 299], "argnum": [2, 165, 299, 481], "argpars": 5, "argpartit": 0, "argsort": 0, "argument": [1, 31, 65, 79, 94, 136, 165, 299, 311, 312, 313, 324, 400, 476, 480, 481, 486, 487, 488], "argumentpars": 5, "ari": [84, 85, 86], "aris": 484, "arm": 8, "arm64": 8, "around": 5, "arr": [0, 262, 482], "arr_0": 486, "arrai": [0, 1, 2, 5, 6, 7, 10, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 115, 116, 117, 118, 121, 122, 125, 126, 127, 128, 129, 130, 131, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 221, 222, 223, 224, 225, 226, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 283, 284, 285, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 309, 324, 328, 349, 358, 365, 368, 373, 379, 400, 401, 402, 403, 404, 405, 406, 407, 408, 414, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 446, 449, 452, 455, 456, 457, 458, 459, 460, 461, 466, 467, 468, 469, 470, 471, 472, 479, 480, 481, 483, 484, 485, 487], "array_equ": [0, 16, 172], "arrayfir": 7, "arxiv": [5, 344, 348, 350, 357, 382, 413, 435, 455, 461], "as_strid": 0, "ascend": [186, 187], "ask": [5, 480], "assert": [1, 2, 143], "assign": [0, 2, 38, 452], "associ": [2, 265, 266, 483], "assum": [0, 2, 5, 90, 185, 186, 187, 190, 311, 324, 326, 327, 344, 355, 356], "astyp": [0, 1, 2, 5, 143, 358, 484], "atleast": 0, "atleast_1d": 0, "atleast_2d": 0, "atleast_3d": 0, "atol": [0, 16, 172], "atom": [1, 143], "atomic_fetch_add_explicit": 1, "atomic_output": [1, 143], "attach": 2, "attempt": 94, "attend": 378, "attent": [146, 363, 378, 391, 399], "attention_norm": 5, "attribut": [1, 9, 10, 11, 30, 315, 371, 452, 474], "audio": 400, "auto": [0, 2, 8], "autom": 481, "automat": [1, 2, 7, 143, 194, 480, 485, 486, 487], "autoregress": 5, "avail": [2, 4, 5, 6, 8, 10, 124, 214, 319, 487], "averag": [326, 327, 455, 456, 458, 459, 460, 480], "avgpool1d": 324, "avgpool2d": 324, "avoid": [1, 2, 370, 479], "awai": [2, 5], "awar": [479, 483], "ax": [0, 2, 15, 17, 26, 27, 79, 112, 138, 148, 149, 151, 152, 154, 155, 157, 158, 159, 171, 189, 203, 205, 207, 221, 232, 235, 259, 273, 278, 280, 284, 285, 291, 295, 300, 481], "axes_a": 0, "axes_b": 0, "axi": [0, 2, 5, 6, 15, 17, 26, 27, 28, 29, 33, 34, 35, 36, 41, 42, 43, 44, 56, 57, 58, 59, 63, 71, 74, 75, 76, 80, 95, 108, 109, 110, 111, 118, 121, 138, 142, 144, 147, 150, 153, 154, 155, 156, 157, 158, 159, 168, 185, 189, 203, 205, 207, 221, 223, 232, 233, 235, 236, 241, 247, 256, 259, 273, 274, 275, 278, 279, 280, 284, 285, 287, 288, 292, 293, 294, 295, 300, 301, 303, 326, 327, 342, 355, 356, 383, 414, 420, 422, 423, 427, 432, 434, 442, 443, 482], "axis1": [0, 46, 77, 118, 285, 294], "axis2": [0, 46, 77, 118, 285, 294], "axpbi": 2, "axpby_": 2, "axpby_gener": 2, "axpby_general_": 2, "axpby_impl": 2, "axpby_impl_acceler": 2, "b": [0, 1, 2, 3, 5, 13, 14, 16, 24, 82, 87, 88, 89, 90, 128, 129, 133, 143, 161, 163, 166, 167, 171, 172, 180, 181, 182, 185, 189, 199, 200, 202, 204, 206, 222, 224, 228, 231, 234, 237, 244, 255, 258, 283, 291, 299, 312, 313, 342, 352, 383, 400, 414, 481, 482, 483, 484, 485, 486, 487], "b1": 163, "b2": 163, "b_": [343, 349], "b_stride": 1, "ba": [458, 460], "back": [5, 112, 214, 484], "backend": [1, 8, 123, 124], "backward": [1, 479, 481], "bad": 483, "balanc": 429, "baltimor": 189, "bandwidth": [479, 480], "bar": 480, "base": [0, 2, 145, 189, 196, 198, 234, 386, 399, 452, 454, 460, 474, 476, 479, 482], "base_idx": 1, "basi": 474, "basic": [4, 260, 481], "batch": [5, 14, 90, 163, 164, 204, 245, 328, 330, 331, 332, 333, 334, 335, 337, 338, 343, 349, 378, 383, 400, 483], "batch_idx": 1, "batch_iter": [6, 454], "batch_siz": [6, 454], "batchnorm": 324, "becaus": [5, 211, 324, 483], "been": [0, 2, 5, 212, 483], "befor": [1, 2, 5, 8, 28, 143, 233, 362, 399, 463, 480, 482, 483], "before_1": 232, "before_2": 232, "before_i": 232, "before_n": 232, "beforehand": 231, "beggin": 259, "begin": [83, 184, 213, 237, 327, 343, 349, 356, 397, 415, 426, 433, 439, 445, 446], "behav": 112, "behavior": [245, 429, 482, 483], "behaviour": [112, 183, 184], "behind": 481, "being": [281, 324], "bell": 2, "below": [2, 8, 189, 296, 298, 317, 400, 483], "bench": 2, "benchmark": [2, 479], "benefici": [337, 338, 483], "best": 480, "beta": [0, 2, 14, 116, 141, 237, 328, 344, 348, 350, 433, 454, 458, 459, 460, 461], "beta_": 2, "beta_1": [456, 458, 459, 460, 461], "beta_2": [458, 459, 460, 461], "better": [481, 487], "between": [0, 2, 7, 93, 159, 399, 422, 425, 426, 429, 470, 480, 483, 484, 487], "beyond": [259, 468, 471], "bfloat16": [2, 11, 168, 317, 484], "bfloat16_t": 2, "bia": [5, 116, 141, 142, 164, 237, 238, 311, 324, 330, 331, 332, 333, 334, 335, 343, 349, 350, 352, 363, 365, 375, 378, 381, 383, 458, 459, 460, 463, 481], "bias": [0, 116, 141, 164, 237, 238, 343, 349, 363, 375, 378], "bicub": 400, "big": [1, 479], "bigger": [5, 456], "bilinear": [1, 400], "binari": [194, 262, 263, 264, 265, 266, 301, 397, 421, 446, 479], "binary_cross_entropi": [324, 479], "bit": [0, 116, 141, 164, 180, 237, 238, 258, 307, 317, 358, 380, 381, 382], "bitwis": [0, 87, 88, 89, 180, 258], "bitwise_and": 0, "bitwise_or": 0, "bitwise_xor": 0, "block": [0, 2, 5, 90, 399], "block_masked_mm": 0, "block_siz": [0, 90], "bn": 328, "bodi": [1, 143], "bool": [0, 1, 2, 15, 16, 17, 26, 27, 33, 34, 35, 36, 41, 42, 43, 44, 56, 57, 58, 59, 63, 75, 76, 78, 80, 82, 94, 101, 108, 109, 110, 111, 123, 124, 143, 145, 164, 172, 178, 183, 184, 189, 192, 194, 203, 205, 207, 208, 214, 217, 221, 235, 238, 280, 284, 300, 328, 330, 331, 332, 333, 334, 335, 343, 344, 348, 349, 350, 352, 358, 362, 363, 365, 370, 372, 375, 378, 381, 383, 386, 391, 399, 400, 421, 424, 456, 467], "bool_": [11, 317], "boolean": [0, 16, 82, 172, 173, 174, 175, 176, 177, 178, 200, 201, 202, 317, 374, 482], "both": [1, 2, 13, 87, 88, 89, 128, 129, 133, 166, 167, 178, 180, 181, 182, 189, 199, 206, 222, 224, 228, 234, 241, 255, 258, 283, 307, 326, 327, 348, 349, 355, 356, 454, 479, 480, 481, 485, 487], "bottom": 400, "bound": [0, 248, 251, 252, 341, 408, 479, 482, 487], "boundari": 470, "bracket": 5, "brain": 317, "break": 484, "bregler": 337, "broadcast": [0, 2, 13, 16, 87, 88, 89, 91, 93, 128, 129, 133, 162, 166, 167, 172, 180, 181, 182, 199, 204, 206, 222, 224, 228, 234, 236, 240, 241, 245, 251, 252, 255, 258, 283, 288, 304, 378], "broadcast_arrai": [0, 2], "broadcast_to": 0, "broadcasted_input": 2, "brought": 7, "btl_tcp_link": 480, "buffer": [1, 2, 211, 484], "bui": 5, "build": [3, 5, 7, 404, 452, 479], "build_ext": [2, 8], "build_shared_lib": [2, 8], "built": [1, 2, 8, 483], "bundl": 5, "byte": [51, 61, 211, 212, 213, 216, 217, 218, 317], "c": [0, 1, 2, 5, 14, 189, 326, 327, 328, 330, 331, 332, 333, 334, 335, 337, 338, 348, 349, 355, 356, 484, 485, 487], "c_": [349, 461], "c_in": [98, 99, 100, 101, 102, 103, 104], "c_j": [326, 327, 355, 356], "c_out": [98, 99, 100, 101, 102, 103, 104], "c_pad": 1, "c_t": [349, 461], "cach": [5, 8, 209, 211, 212, 216, 479], "calcul": [189, 421, 424, 430, 456], "call": [2, 3, 5, 6, 31, 126, 161, 209, 213, 324, 340, 363, 375, 380, 388, 452, 454, 463, 479, 480, 481, 483], "callabl": [94, 112, 143, 165, 179, 299, 302, 303, 307, 308, 310, 311, 312, 313, 358, 359, 362, 370, 383, 388, 399, 401, 402, 403, 404, 405, 406, 407, 408, 455, 456, 457, 458, 459, 460, 461, 466, 467, 468, 469, 470, 471, 472], "can": [1, 2, 3, 5, 7, 8, 13, 18, 65, 79, 83, 87, 88, 89, 94, 118, 119, 120, 128, 129, 133, 136, 166, 167, 180, 181, 182, 189, 199, 206, 218, 222, 224, 228, 234, 240, 241, 248, 251, 252, 255, 258, 263, 283, 294, 299, 313, 324, 327, 340, 341, 356, 362, 375, 380, 388, 400, 423, 449, 452, 454, 462, 463, 476, 478, 479, 480, 481, 482, 483, 484, 485, 486, 487, 488], "cannot": [5, 93, 482, 484], "captur": [2, 3, 94, 219, 220, 324, 479], "care": [5, 483], "carri": 2, "cartesian": 208, "case": [2, 5, 121, 122, 123, 125, 126, 127, 149, 152, 153, 155, 156, 157, 158, 159, 183, 184, 185, 186, 187, 188, 190, 191, 192, 204, 257, 278, 327, 337, 356, 397, 415, 433, 439, 445, 446, 462, 463, 479, 481, 485, 486, 487, 488], "cast": [2, 37, 156, 157, 158, 194, 358, 370, 484], "caster": 2, "categor": 5, "categori": [11, 178, 317], "catlas_saxpbi": 2, "caus": [324, 479, 483], "causal": 5, "caution": 83, "cd": [3, 8], "cdf": [242, 341, 411], "cdot": [413, 422, 425, 441], "ceil": 0, "ceildiv": 1, "cell": 349, "celu": 324, "certain": [2, 372, 479], "chang": [83, 94, 268, 301, 376, 381, 400, 426, 433, 479, 484], "channel": [1, 98, 99, 100, 101, 102, 103, 104, 328, 330, 331, 332, 333, 334, 335, 337, 338], "channel_idx": 1, "charact": 310, "check": [0, 2, 8, 82, 124, 178, 186, 187, 214, 365, 481, 482], "checklist": 480, "checkout": [3, 479], "checkpoint": [399, 454], "chen": 461, "child": 377, "children": 324, "chip": 8, "choleski": 184, "choos": [5, 145, 386], "chosen": 131, "clamp": 159, "clang": 8, "clariti": 481, "class": [2, 5, 6, 9, 10, 11, 30, 112, 120, 315, 325, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350, 351, 352, 353, 354, 355, 356, 357, 378, 379, 380, 381, 382, 383, 384, 385, 386, 387, 388, 389, 390, 391, 392, 393, 394, 395, 396, 397, 398, 399, 400, 409, 410, 411, 412, 413, 414, 415, 416, 417, 418, 419, 420, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445, 446, 447, 452, 455, 456, 457, 458, 459, 460, 461, 466, 467, 474], "class_pred": 307, "classif": [404, 405], "classifi": 6, "classmethod": [380, 381], "clear": 209, "click": 8, "clip": [0, 309, 421, 456], "clip_threshold": 456, "clipped_grad": 309, "clone": 8, "close": [4, 7, 8, 16, 172], "closer": 311, "cmake": [3, 8], "cmake_arg": 3, "cmake_build_parallel_level": 8, "cmake_build_typ": 8, "cmake_current_list_dir": 2, "cmake_host_system_processor": 8, "cmake_library_output_directori": 2, "cmakebuild": 2, "cmakeextens": 2, "cmakelist": 2, "cmdclass": 2, "co": [0, 2, 112, 391, 481], "code": [1, 143, 479, 480, 483], "coeffici": [2, 455, 456, 458, 459, 460, 461], "col": 296, "col_contigu": 2, "cold": 8, "collect": [2, 311, 312, 478], "column": [2, 140, 169, 186, 237], "com": [8, 480], "combin": [5, 191, 313], "come": [2, 5, 480, 481], "command": [2, 3, 8, 480], "command_buff": 2, "common": [2, 454, 479, 483], "commonli": [6, 376, 449, 479], "commun": [7, 120, 123, 124], "compar": [2, 82, 479], "comparison": [16, 133, 166, 167, 181, 182, 228], "compat": [5, 241, 245, 341, 486], "compil": [0, 3, 7, 8, 119, 132, 143, 480, 481, 483], "compiled_fun": 479, "compiled_grad_fn": 479, "complet": [4, 5, 8, 217, 376, 377, 481, 487], "complex": [2, 96, 97, 154, 155, 156, 157, 158, 170, 186, 187, 253, 310, 317, 324, 377, 479, 481], "complex64": [2, 11, 317], "complex64_t": 2, "complexflo": 11, "compon": [2, 5], "compos": [7, 324, 479, 481, 485], "composit": 485, "compress": 266, "compromis": 5, "comput": [0, 1, 2, 4, 5, 6, 7, 8, 108, 109, 110, 111, 112, 116, 131, 139, 141, 145, 165, 179, 183, 184, 185, 186, 187, 188, 189, 192, 199, 207, 231, 237, 255, 273, 280, 281, 291, 299, 300, 302, 308, 324, 328, 343, 344, 348, 349, 350, 363, 376, 381, 382, 386, 399, 402, 403, 404, 405, 412, 413, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 454, 455, 456, 458, 459, 460, 461, 465, 479, 480, 481, 485, 487], "computation": 483, "compute_encod": 2, "concaten": [0, 5, 121], "concept": 452, "concis": 5, "concret": [2, 343, 349, 352, 383, 483, 487], "conda": [8, 480], "condit": [0, 2, 304, 487], "config": [2, 480], "configu": 454, "configur": [116, 141, 480], "confirm": 480, "confus": 6, "conj": 97, "conjug": [0, 96], "connect": 480, "consecut": [145, 237, 386], "consequ": 5, "consid": [5, 16, 82, 172, 310, 311, 312, 344, 478], "consider": 479, "const": [0, 1, 2, 424], "constant": [0, 2, 5, 8, 142, 144, 232, 324, 328, 344, 350, 382, 424, 434, 466, 468, 479, 484], "constant_valu": 232, "constitut": 311, "construct": [0, 2, 6, 45, 117, 162, 229, 292, 305], "consum": 483, "contain": [2, 5, 8, 28, 29, 68, 94, 118, 131, 153, 154, 155, 163, 164, 186, 189, 200, 201, 202, 237, 275, 304, 309, 324, 362, 364, 365, 371, 399, 430, 449, 452, 479, 480, 481], "content": [8, 362, 479], "context": 282, "contigu": [1, 2, 83, 143], "continu": [329, 409, 481], "contract": [0, 131], "contrast": 459, "contribut": 2, "contriv": [481, 487], "control": [0, 351, 476, 483], "conv": 105, "conv1d": [0, 324], "conv2d": [0, 324], "conv3d": [0, 324], "conv_gener": 0, "conv_transpose1d": 0, "conv_transpose2d": 0, "conv_transpose3d": 0, "conveni": [1, 2, 6, 178], "convent": [18, 105, 130, 131, 400, 459], "convers": 7, "convert": [0, 1, 2, 78, 84, 85, 86, 115, 159, 239, 380, 381, 483, 484, 485], "convolut": [0, 98, 99, 100, 101, 102, 103, 104, 105, 330, 331, 332, 333, 334, 335, 337, 338], "convolv": [98, 99, 100, 101, 102, 103, 104], "convtranspose1d": 324, "convtranspose2d": 324, "convtranspose3d": 324, "coordin": [0, 208], "copi": [0, 1, 2, 5, 7, 233, 274, 484], "copy_inplac": 2, "copytyp": 2, "core": [1, 2, 3, 4, 5, 6, 307, 324, 326, 327, 328, 348, 355, 356, 365, 368, 370, 373, 400, 401, 402, 403, 404, 405, 406, 407, 408, 421, 423, 430, 449, 452, 454, 479, 480, 484, 485], "corner": 400, "correct": [2, 8, 458, 459, 460, 482, 483], "correctli": 38, "correl": [101, 337], "correspond": [0, 2, 15, 17, 78, 93, 116, 118, 141, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 186, 203, 205, 221, 235, 284, 291, 303, 311, 481], "cos_first": 391, "cosh": [0, 429], "cosin": [0, 19, 20, 106, 107, 422, 468, 470, 481], "cosine_decai": [454, 470], "cosine_similarity_loss": 324, "cost": [8, 456, 480, 483], "costli": 483, "cot": 1, "cot_index": 1, "cotan": [2, 112], "cotang": [1, 2, 112, 302], "could": [5, 324], "count": [324, 470], "counter": 476, "cours": 481, "coursera": 466, "cov": 245, "covari": [245, 328], "cover": 2, "cpp": 2, "cpu": [7, 8, 186, 187, 190, 487], "cpython": 2, "crash": [83, 479], "creat": [0, 2, 5, 8, 83, 123, 140, 169, 282, 324, 452, 454, 470, 479, 482, 484], "create_additive_causal_mask": 5, "criteria": 2, "cross": [6, 101, 421, 423], "cross_entropi": [6, 324], "crowd": 5, "cry": 5, "cubic": 400, "cummax": 0, "cummin": 0, "cumprod": 0, "cumsum": 0, "cumul": [0, 83, 108, 109, 110, 111], "current": [5, 7, 8, 83, 90, 100, 103, 104, 127, 210, 212, 237, 313, 324, 456, 480, 483], "custom": [7, 112, 143, 399], "custom_decod": 399, "custom_encod": 399, "custom_funct": 1, "custom_kernel_myexp_float": 1, "custom_tim": 2, "cvpr": 337, "cycl": 478, "d": [0, 1, 2, 5, 100, 104, 117, 118, 171, 189, 204, 208, 231, 287, 294, 296, 297, 298, 314, 332, 335, 338, 343, 349, 383, 455, 458, 460, 487], "d1": 487, "d2": 487, "d2fdx2": 481, "d_i": 352, "dampen": 467, "darwin": 2, "data": [0, 2, 6, 7, 10, 18, 125, 140, 156, 157, 162, 169, 193, 225, 229, 242, 251, 294, 296, 301, 305, 338, 401, 402, 403, 404, 405, 406, 407, 408, 479, 480, 482, 484], "dataset": [4, 480, 483], "datatyp": 51, "dbuild_shared_lib": 8, "dcmake_build_typ": 8, "ddof": [0, 75, 80, 280, 300], "deal": 479, "debug": [1, 3, 480], "debugg": 7, "decai": [456, 459, 461, 467, 468, 469, 472], "decay_r": [456, 469, 472], "decay_step": 468, "decent": 6, "decid": [311, 362], "decim": [0, 66, 260], "declar": 2, "decltyp": 1, "decod": 399, "decomposit": [183, 184, 191], "decor": [1, 112], "decoupl": 459, "deep": [328, 402, 403, 404, 405], "def": [1, 2, 4, 5, 6, 112, 143, 299, 324, 452, 479, 480, 481, 482, 483, 484, 487], "default": [1, 2, 8, 14, 15, 16, 17, 18, 26, 27, 28, 29, 82, 83, 90, 94, 95, 98, 99, 100, 101, 102, 103, 104, 112, 113, 114, 116, 117, 118, 121, 122, 123, 125, 126, 127, 140, 141, 143, 145, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 163, 164, 165, 168, 169, 172, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 203, 205, 207, 208, 216, 217, 218, 221, 225, 229, 232, 233, 235, 237, 238, 240, 241, 242, 244, 245, 246, 247, 248, 250, 251, 252, 256, 257, 260, 267, 268, 274, 275, 278, 279, 280, 282, 284, 286, 291, 293, 294, 295, 296, 297, 298, 299, 300, 303, 305, 307, 317, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 339, 342, 343, 345, 348, 349, 351, 352, 355, 356, 358, 363, 365, 370, 372, 375, 378, 379, 380, 381, 383, 386, 391, 395, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 414, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 452, 455, 456, 457, 458, 459, 460, 461, 466, 467, 468, 476, 478, 479, 481, 484, 486, 488], "default_devic": 488, "default_stream": 488, "defin": [1, 2, 4, 5, 6, 8, 112, 126, 143, 164, 185, 189, 238, 307, 310, 484], "definit": [112, 183, 184, 245], "degre": [0, 239, 434], "delta": [426, 455], "delv": [404, 405], "demonstr": 484, "denomin": [348, 422, 455, 457, 458, 459, 460, 466], "dens": [208, 487], "depend": [0, 2, 3, 4, 8, 78, 189, 343, 349, 383, 480, 482, 486, 487], "depth": [310, 332, 335, 338, 481], "dequant": [0, 237], "deriv": [2, 481, 483], "descend": 360, "descent": [467, 479, 483], "describ": [2, 483], "descript": [2, 5, 317], "design": [1, 4, 7, 476, 487], "destin": [0, 2, 60, 127, 223, 236], "destroi": 479, "detach": 481, "detail": [1, 2, 10, 216, 324, 337, 386, 391, 400, 402, 403, 404, 405, 455, 457, 458, 460, 461, 482, 485], "determin": [0, 2, 118, 245, 313, 317, 369, 486], "dev": [2, 8], "develop": [2, 8], "developer_dir": 8, "deviat": [0, 246, 280, 402, 404, 407], "deviatoin": 0, "devic": [1, 2, 7, 8, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 113, 114, 115, 116, 117, 118, 121, 122, 125, 126, 127, 128, 129, 130, 133, 134, 135, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 210, 217, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 267, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 300, 301, 304, 305, 306, 315, 487, 488], "device_info": 218, "devicetyp": 9, "df": 484, "dfdx": [481, 482], "dft": [147, 148, 149, 150, 151, 152, 156, 157, 158], "dhwc": 338, "diag": [0, 191], "diagon": [0, 45, 117, 140, 294, 296, 297, 298], "dict": [94, 136, 194, 210, 263, 264, 265, 309, 368, 373, 376, 377, 452, 454, 462, 463, 465, 478, 481, 486], "dict_kei": [311, 463], "dictionari": [5, 94, 194, 210, 263, 264, 309, 310, 313, 324, 362, 371, 376, 377, 464, 478, 486], "did": 5, "diff": 2, "differ": [7, 178, 283, 301, 433, 481], "differenti": [1, 2, 7, 329, 409], "difficult": 481, "difficulti": [402, 403], "dilat": [0, 98, 99, 100, 101, 102, 103, 104, 330, 331, 333, 334], "dim": [1, 5, 145, 146, 340, 344, 348, 350, 378, 380, 382, 386, 391, 399], "dimens": [0, 2, 5, 15, 17, 26, 27, 62, 68, 78, 84, 85, 86, 94, 99, 100, 101, 103, 104, 118, 138, 145, 154, 155, 157, 158, 159, 163, 164, 171, 183, 184, 186, 187, 188, 189, 190, 191, 192, 203, 204, 205, 207, 221, 235, 236, 237, 241, 250, 280, 284, 288, 291, 295, 300, 328, 330, 331, 332, 333, 334, 335, 337, 338, 342, 343, 344, 348, 349, 350, 378, 382, 383, 386, 399, 400, 414, 423, 479, 481], "dimension": [30, 142, 144, 147, 148, 149, 150, 151, 152, 156, 157, 158, 326, 327, 328, 330, 331, 332, 333, 334, 335, 340, 352, 355, 356, 380, 381, 391, 482, 484], "direct": [2, 5, 360, 461, 487], "directli": [2, 5, 83], "directori": [2, 5, 8], "disabl": [119, 216, 479], "disable_compil": 479, "disappoint": 5, "discard": [5, 310], "discov": 8, "discoveri": 461, "discret": [105, 147, 148, 149, 150, 151, 152, 156, 157, 158, 340, 380], "discuss": 2, "disk": 5, "dispatch": 2, "dispatchthread": 2, "displai": 324, "distanc": [5, 434], "distribut": [7, 8, 240, 241, 242, 244, 245, 246, 251, 252, 352, 402, 403, 404, 405, 407, 408, 424, 427, 432, 434, 449], "diverg": 427, "divid": [0, 2, 38, 161, 237, 255, 480], "divis": [0, 128, 161, 237, 255], "divisor": [280, 300], "divmod": 0, "dloss_dw": 481, "dloss_dx": 481, "dlpack": 484, "dlvalu": 299, "dmlx_build_cpu": 8, "dmlx_build_gguf": 8, "dmlx_build_safetensor": 8, "dmlx_metal_debug": 3, "dmlx_metal_jit": 8, "do": [0, 2, 5, 8, 301, 324, 364, 375, 449, 452, 459, 479, 480, 481, 483], "doc": [2, 6, 480], "document": [2, 3, 65, 79, 263, 264, 317, 479, 481, 482], "doe": [0, 2, 3, 5, 8, 211, 301, 309, 324, 479, 482, 483, 484], "doesn": [2, 324], "domain": [251, 480], "don": [1, 8, 479, 487], "done": [324, 336, 382, 479, 480, 483, 484], "dot": [183, 188, 192, 291, 310, 367, 378], "doubl": [0, 5], "doubt": 5, "down": [5, 309], "dparam": 299, "draw": 241, "drop": 362, "dropout": [324, 337, 338, 372, 399, 479], "dropout2d": 324, "dropout3d": 324, "dst": 127, "dt": 134, "dtype": [0, 1, 2, 5, 11, 18, 30, 37, 38, 78, 81, 125, 126, 140, 143, 159, 162, 169, 178, 186, 187, 189, 190, 193, 229, 242, 244, 245, 246, 248, 251, 252, 294, 296, 301, 305, 317, 370, 400, 401, 402, 403, 404, 405, 406, 407, 408, 421, 423, 430, 468, 469, 470, 471, 472, 479, 480, 481, 482, 484, 485, 486], "dtypecategori": [178, 317], "dual": 429, "duchi": 457, "dure": [3, 94, 336, 337, 338, 400, 484], "dx": 112, "dy": 112, "dyld": 480, "dyld_library_path": 480, "dylib": 2, "dynam": 483, "e": [2, 6, 8, 112, 134, 143, 163, 164, 179, 269, 328, 330, 331, 332, 333, 334, 335, 337, 338, 344, 348, 350, 363, 382, 419, 420, 442, 443, 448, 454, 457, 479, 483, 488], "e5": 317, "e8": 317, "each": [0, 1, 2, 68, 116, 136, 141, 145, 164, 178, 183, 184, 186, 187, 188, 191, 192, 204, 208, 232, 237, 238, 241, 256, 265, 266, 275, 292, 295, 301, 303, 304, 337, 338, 340, 343, 344, 349, 383, 386, 399, 421, 423, 476, 479, 480, 483], "eager": 483, "earli": 337, "earlier": 2, "eas": 5, "easi": [2, 324, 480], "easier": [1, 483], "edg": [93, 232, 400, 479], "edit": [8, 377], "effect": [337, 479, 483], "effici": [5, 7, 163, 337, 386, 483, 485], "eigenvalu": [186, 187], "eigenvector": 186, "einstein": [130, 131], "einsum": 131, "either": [8, 13, 65, 78, 79, 87, 88, 89, 93, 128, 129, 133, 161, 166, 167, 180, 181, 182, 189, 199, 204, 206, 222, 224, 228, 234, 255, 258, 283, 299, 327, 356, 388, 400, 404, 405, 484], "elem": [1, 143], "elem_to_loc": [1, 2], "element": [0, 1, 2, 12, 13, 19, 20, 21, 22, 23, 24, 25, 28, 70, 83, 87, 88, 89, 92, 106, 107, 108, 109, 110, 111, 116, 128, 129, 133, 134, 135, 137, 139, 140, 141, 160, 161, 164, 166, 167, 172, 173, 174, 175, 176, 177, 180, 181, 182, 195, 196, 197, 198, 199, 200, 201, 202, 206, 208, 222, 224, 226, 228, 233, 234, 237, 238, 254, 255, 256, 258, 259, 261, 269, 270, 271, 272, 276, 277, 283, 287, 289, 290, 293, 299, 301, 304, 329, 336, 337, 338, 343, 347, 349, 357, 379, 383, 386, 390, 409, 416, 417, 419, 420, 435, 436, 438, 441, 442, 443, 444, 479, 481], "elementwis": [1, 96, 97], "elif": 5, "ellipsi": 482, "elman": 383, "els": [0, 2, 5, 324, 363, 480, 483], "elsewher": [296, 482], "elu": [324, 439], "emb": [5, 340, 380, 391], "embed": [5, 307, 324, 380, 386, 391, 422], "empti": [127, 245], "enabl": [3, 5, 8, 94, 132, 467], "encod": [2, 145, 386, 391, 399, 423], "encount": [2, 481], "end": [118, 184, 214, 237, 259, 327, 343, 349, 356, 397, 415, 426, 433, 439, 445, 446, 468, 471], "end_axi": [0, 49, 159], "end_encod": 2, "endif": 2, "endswith": 363, "enhanc": [5, 386, 483], "enjoi": 2, "enough": [2, 483], "ensur": [0, 1, 2, 8, 143, 309, 429, 480], "ensure_row_contigu": [1, 143], "enter": 5, "entir": [15, 17, 26, 27, 203, 205, 207, 221, 235, 280, 284, 300, 337, 338], "entri": [0, 247, 337, 338], "entropi": [6, 421, 423], "enumer": 324, "environ": [8, 119, 132, 480], "ep": [4, 142, 144, 328, 344, 348, 350, 382, 422, 424, 434, 454, 455, 456, 457, 458, 459, 460, 466], "epoch": 6, "epsilon": [328, 344, 348, 350, 382, 422, 424, 455, 457, 458, 459, 460, 466], "epsilon_1": 456, "epsilon_2": 456, "equal": [0, 16, 28, 82, 140, 167, 172, 182, 228, 233, 248, 275, 348, 352], "equal_nan": [0, 16, 82, 172], "equat": [130, 131], "equival": [0, 2, 31, 65, 79, 126, 129, 161, 164, 168, 287, 329, 339, 341, 345, 346, 347, 353, 354, 377, 379, 381, 384, 385, 387, 389, 392, 393, 394, 395, 396, 398], "erf": [0, 135, 479], "erfinv": 0, "error": [0, 2, 8, 123, 134, 135, 217, 218, 275, 341, 411, 412, 413, 429, 431, 481, 484], "error_norm": 4, "estim": [458, 460], "eta": 461, "etc": [2, 237, 324, 400, 480], "eval": [2, 3, 4, 5, 6, 324, 452, 454, 479, 480, 481, 483, 485], "eval_cpu": 2, "eval_fn": 6, "eval_gpu": 2, "evalu": [2, 5, 6, 7, 127, 136, 179, 302, 324, 361, 372, 452, 454, 479, 485], "even": [1, 2, 5, 94, 479, 483, 484], "evenli": [0, 193], "everi": [237, 311, 454, 472, 481], "everyth": [5, 480], "everywher": 0, "exact": [412, 413], "exactli": [2, 5, 145, 365, 481], "exampl": [0, 3, 4, 5, 6, 8, 18, 38, 112, 143, 159, 178, 186, 187, 189, 190, 282, 287, 309, 312, 313, 324, 326, 327, 328, 348, 355, 356, 363, 365, 372, 375, 400, 401, 402, 403, 404, 405, 406, 407, 408, 421, 423, 430, 449, 454, 463, 468, 469, 470, 471, 472, 476, 481, 482, 483, 484, 485, 486], "exce": 309, "exceed": 217, "except": [7, 140, 153, 154, 156, 157, 158, 344, 365, 482, 484], "exclud": [236, 288], "exclus": [0, 83, 89], "execut": [2, 8, 84, 85, 86, 213, 484, 487], "exist": [2, 3, 5, 363, 375], "exp": [0, 1, 139, 143, 199, 203, 242, 273, 329, 339, 390, 409, 410, 427, 439, 440, 444, 479, 487], "exp_elementwis": [1, 143], "expand_dim": 0, "expect": [2, 5, 330, 331, 332, 333, 334, 335, 336, 337, 338, 391, 399, 424, 479, 482], "expens": 399, "expensive_fun": 483, "experiment": 484, "explain": 2, "explicit": [2, 463, 476, 484], "explicitli": [163, 324, 476], "explor": 8, "expm1": 0, "exponenti": [0, 137, 139, 329, 339, 387, 409, 410, 439, 469], "exponential_decai": 454, "export": 8, "ext_modul": 2, "extend": [2, 232], "extens": [7, 194, 219, 369, 486], "extern": 484, "extra": [1, 311, 312], "extract": [0, 5, 45, 117, 118, 324, 362, 452, 480], "extras_requir": 2, "extrem": [482, 483], "ey": [0, 5, 188, 192], "f": [0, 2, 4, 6, 112, 189, 324, 349, 459, 479, 484], "f_jvp": 112, "f_t": 349, "f_vjp": 112, "f_vmap": 112, "face": 5, "factor": [2, 14, 168, 183, 184, 190, 400, 423, 469, 472], "fall": [2, 112], "fallback": 2, "fals": [0, 1, 2, 5, 15, 16, 17, 26, 27, 33, 34, 35, 36, 41, 42, 43, 44, 56, 57, 58, 59, 63, 75, 76, 80, 82, 94, 101, 108, 109, 110, 111, 123, 143, 172, 178, 183, 184, 189, 192, 194, 203, 205, 207, 208, 217, 221, 235, 280, 284, 300, 304, 307, 310, 311, 312, 313, 317, 344, 348, 350, 352, 363, 365, 375, 378, 381, 386, 391, 399, 400, 421, 424, 456, 467, 484], "famili": 5, "fan": [402, 403, 404, 405], "fan_in": [402, 403, 404, 405], "fan_out": [402, 403, 404, 405], "far": 454, "fast": [1, 7, 341, 413, 480, 487], "faster": [1, 2, 8, 129, 411, 421, 479, 481], "featur": [1, 7, 98, 99, 100, 101, 102, 103, 104, 145, 328, 343, 344, 348, 349, 350, 352, 381, 382, 383, 386, 399, 400, 479, 483], "feed": 5, "feed_forward": 5, "feedforward": [402, 403], "feel": 5, "fetch": 1, "few": [1, 2, 5, 6, 7, 8, 480, 483, 485], "ffn": 5, "ffn_norm": 5, "fft": 7, "figur": 480, "file": [5, 8, 194, 262, 263, 264, 265, 266, 365, 369, 480, 481, 486], "file_or_weight": 365, "fill": [0, 2, 162, 230, 296, 306, 401, 402, 403, 404, 405, 407, 408], "filter": [0, 105, 330, 331, 332, 333, 334, 335, 358, 362], "filter_and_map": 324, "filter_fn": [358, 362], "final": [2, 4, 5, 6, 168, 468, 471], "find": [2, 4, 8, 480], "find_packag": 2, "finder": 8, "fine": [476, 483], "finetun": 324, "finish": 2, "finit": [0, 173, 225], "first": [0, 1, 2, 3, 4, 5, 6, 8, 118, 121, 159, 165, 178, 180, 191, 200, 202, 204, 233, 250, 258, 285, 291, 294, 299, 310, 312, 313, 324, 327, 344, 356, 400, 422, 430, 456, 458, 459, 460, 463, 479, 481, 484, 487], "first_lay": 483, "fit": [2, 237, 487], "five": 479, "fix": [2, 5, 8, 483], "flag": [2, 8, 479, 484], "flat": [163, 164, 310, 314], "flat_param": 265, "flatten": [0, 28, 29, 108, 109, 110, 111, 189, 231, 233, 236, 256, 259, 274, 287, 288, 293, 310], "flexibl": 7, "flexibli": 377, "flip": [0, 101, 105], "float": [0, 1, 2, 11, 14, 16, 18, 78, 142, 143, 144, 145, 146, 161, 162, 168, 172, 178, 189, 225, 238, 240, 244, 246, 309, 317, 328, 336, 337, 338, 344, 348, 350, 358, 370, 382, 386, 391, 397, 399, 400, 401, 402, 403, 404, 405, 407, 408, 422, 423, 424, 426, 430, 433, 434, 445, 446, 455, 456, 457, 458, 459, 460, 461, 466, 467, 468, 469, 471, 472], "float16": [1, 2, 11, 143, 168, 194, 317, 358, 483, 484], "float16_t": [1, 2], "float32": [0, 1, 2, 11, 18, 140, 143, 146, 168, 169, 178, 186, 187, 189, 190, 193, 229, 242, 244, 245, 246, 251, 252, 296, 305, 317, 400, 401, 402, 403, 404, 405, 406, 407, 408, 421, 423, 430, 468, 469, 470, 471, 472, 479, 480, 481, 482, 483, 484, 485, 486], "float64": 178, "floor": [0, 1, 161], "floor_divid": 0, "flow": [0, 281, 483], "flush": 2, "fn": [308, 311, 312, 313, 485], "follow": [1, 2, 5, 6, 7, 8, 18, 105, 116, 141, 163, 189, 232, 237, 312, 324, 412, 413, 427, 455, 456, 457, 458, 459, 460, 461, 467, 476, 479, 480, 481, 487], "foo": 480, "food": 5, "forc": [5, 6, 324, 480, 485], "forg": 8, "formal": [116, 141, 237], "format": [5, 194, 262, 263, 264, 265, 266, 484], "formul": [329, 339], "formula": 433, "forth": 400, "forward": [1, 2, 299, 479, 483], "found": 362, "four": 328, "fourier": [147, 148, 149, 150, 151, 152, 156, 157, 158], "frac": [134, 237, 269, 326, 327, 328, 336, 337, 338, 344, 348, 350, 352, 355, 356, 382, 390, 402, 403, 404, 405, 422, 424, 426, 429, 440, 442, 443, 455, 457, 458, 459, 460, 466], "fraction": 18, "framework": [2, 7], "free": 216, "freez": [324, 375, 452], "freq": 145, "frequenc": [145, 386, 391], "frequent": [479, 483], "friend": 5, "fro": 189, "frobeniu": 189, "from": [0, 1, 2, 5, 6, 7, 83, 115, 116, 118, 121, 122, 125, 126, 127, 141, 143, 154, 155, 157, 158, 162, 163, 168, 189, 194, 204, 208, 213, 216, 230, 237, 239, 240, 241, 242, 243, 244, 248, 251, 265, 278, 281, 283, 287, 288, 293, 294, 304, 306, 310, 311, 312, 313, 314, 324, 352, 363, 365, 378, 402, 403, 404, 405, 407, 408, 424, 433, 449, 454, 478, 479, 480, 481, 483, 484, 485, 486, 487], "from_embed": 380, "from_linear": 381, "front": 2, "frozen": [324, 363, 373, 375, 381, 452], "fuction": 129, "full": [0, 1, 2, 6, 65, 79, 105, 273, 376, 377, 424, 479, 480, 483], "full_turn": 391, "fulli": [2, 7, 480, 484, 487], "fun": [94, 165, 179, 299, 302, 303, 479, 482, 483, 487], "fun1": 483, "func": 383, "function": [0, 1, 2, 3, 4, 5, 6, 7, 16, 18, 83, 94, 112, 129, 134, 135, 143, 165, 172, 179, 183, 184, 186, 187, 188, 189, 190, 191, 192, 204, 218, 269, 299, 302, 303, 308, 309, 311, 312, 313, 324, 329, 339, 341, 342, 345, 346, 347, 353, 354, 357, 359, 363, 370, 375, 379, 383, 384, 385, 387, 388, 389, 390, 392, 393, 394, 395, 396, 397, 398, 399, 411, 412, 413, 414, 415, 416, 417, 419, 420, 421, 435, 440, 442, 443, 444, 445, 446, 447, 449, 454, 463, 476, 478, 480, 482, 483, 484, 486], "functool": 479, "further": [2, 8, 481], "fuse": [1, 479], "fusibl": 479, "futur": [5, 381, 482, 483], "g": [3, 8, 112, 143, 189, 237, 349, 448, 466, 467, 483, 488], "g_t": [349, 455, 457, 458, 459, 460, 461, 466, 467], "gain": [402, 403, 404, 405], "gamma": [328, 344, 348, 350, 382, 402, 403, 404, 405], "gap": 1, "gate": [342, 343, 414], "gather": [0, 121, 163, 164], "gather_mm": [0, 164], "gather_qmm": 0, "gaurante": 301, "gaussian": [4, 341, 411, 412, 413, 424], "gaussian_nll_loss": 324, "gelu": [324, 412, 413, 479], "gelu_approx": [324, 341, 411], "gelu_fast_approx": [324, 341, 411], "geluapprox": 341, "gelufast": 341, "gener": [0, 1, 2, 3, 4, 11, 18, 101, 140, 143, 154, 155, 193, 208, 240, 245, 246, 247, 248, 251, 252, 399, 476, 479, 482, 483, 488], "general_": 2, "generate_stub": 8, "geq": [397, 446], "get": [2, 4, 6, 8, 99, 100, 101, 103, 104, 113, 114, 210, 211, 212, 213, 243, 324, 479, 481, 483, 487], "get_cache_memori": 209, "get_command_encod": 2, "get_kernel": 2, "gguf": [8, 194, 263, 486], "gh": 1, "gii": 1, "git": 8, "github": [4, 6, 8, 479], "give": [2, 5, 6, 28, 479], "given": [0, 2, 8, 15, 17, 28, 38, 83, 91, 93, 95, 108, 109, 110, 111, 116, 118, 131, 136, 138, 141, 147, 148, 149, 150, 151, 152, 156, 157, 158, 162, 163, 189, 203, 205, 207, 216, 221, 225, 227, 235, 245, 247, 248, 259, 260, 268, 273, 275, 280, 284, 286, 292, 293, 294, 296, 297, 298, 300, 315, 326, 327, 336, 355, 356, 362, 378, 422, 424, 430], "gix": 1, "gix_mult": 1, "giy_mult": 1, "global": [119, 121, 122, 123, 125, 126, 127, 132, 249, 309, 476, 479], "glorot": [402, 403], "glorot_norm": 324, "glorot_uniform": 324, "glu": [5, 324], "gm": 1, "gn": 1, "go": [2, 5, 481], "golub": 189, "good": [2, 8, 454, 479, 480, 487], "goroshin": 337, "gower": 5, "gpu": [1, 3, 7, 8, 210, 482, 487], "gputrac": [3, 219], "grad": [2, 4, 6, 299, 309, 454, 462, 479, 480, 481, 482, 483, 485], "grad_fn": [4, 479, 481], "gradient": [0, 4, 6, 112, 165, 281, 299, 308, 309, 324, 363, 376, 381, 399, 429, 452, 454, 455, 456, 458, 459, 460, 461, 462, 465, 467, 479, 480, 481, 482, 483, 484, 485], "grain": 476, "graph": [2, 5, 6, 7, 481], "great": 3, "greater": [0, 5, 28, 139, 167, 233, 309, 397, 446], "greater_equ": 0, "grep": 8, "grid": [2, 143, 208], "grid_dim": 2, "grid_grad": 1, "grid_idx": 1, "grid_sampl": 1, "grid_sample_grad": 1, "grid_sample_ref": 1, "grid_sample_vjp": 1, "grid_shap": 1, "grid_siz": 1, "ground": [4, 5, 423, 433], "group": [0, 1, 98, 99, 100, 101, 102, 103, 104, 116, 121, 122, 123, 125, 126, 127, 141, 146, 164, 237, 238, 301, 307, 330, 344, 380, 381, 480], "group_dim": 2, "group_siz": [0, 116, 141, 164, 237, 238, 307, 380, 381], "groupnorm": 324, "grow": 483, "gru": 324, "guid": [2, 7], "gw": 1, "h": [1, 2, 98, 99, 100, 102, 103, 104, 189, 327, 328, 331, 332, 334, 335, 337, 338, 343, 349, 356, 383, 481, 483], "h_": [327, 343, 349, 356, 383], "h_in": 1, "h_stride": 1, "h_t": [343, 349, 383], "ha": [2, 3, 5, 6, 7, 8, 78, 94, 118, 127, 153, 154, 156, 157, 158, 165, 183, 184, 186, 187, 188, 191, 192, 208, 212, 241, 328, 343, 349, 352, 383, 452, 454, 479, 482, 483, 485, 487], "had": 5, "hadamard": [0, 168], "hadamard_transform": 0, "half": [2, 18, 248, 252, 386, 483], "halv": [342, 414], "hand": [5, 481, 483], "handi": 481, "handl": [2, 324, 479], "happen": [2, 5, 142, 399, 454, 479, 483], "happi": 5, "hard": 5, "hard_shrink": [324, 345], "hard_tanh": [324, 346], "hardshrink": [324, 415], "hardswish": 324, "hardtanh": [324, 416], "hat": [116, 141, 237], "have": [0, 1, 2, 5, 8, 16, 82, 84, 85, 86, 90, 121, 154, 155, 157, 158, 164, 172, 204, 219, 241, 301, 310, 349, 378, 388, 461, 463, 478, 479, 480, 482, 483, 487], "haven": 5, "hazan": 457, "he": [5, 404, 405], "he_norm": 324, "he_uniform": 324, "head": [146, 378, 399], "header": [2, 143], "heart": 5, "heavi": 5, "height": [327, 328, 331, 332, 334, 335, 337, 338, 356], "hello": [310, 314], "help": [2, 5, 479, 487], "helper": [5, 143, 479], "henc": [0, 2, 237, 479], "hendryck": 413, "here": [2, 5, 454, 479, 481, 483, 486, 487], "hermitian": [186, 187], "hf": 349, "hg": 349, "hh": 383, "hi": [5, 349], "hidden": [343, 349, 383, 399], "hidden_dim": [6, 452, 454], "hidden_s": [343, 349, 383], "hierarchi": 317, "high": [248, 252, 324, 340, 408, 449], "high_pad_s": 0, "higher": [2, 171, 218, 430, 481], "highli": 8, "him": 5, "hing": 425, "hinge_loss": 324, "hinton": 466, "hit": 2, "hn": 343, "ho": 349, "hold": [2, 5, 10, 11, 189, 479], "homebrew": 480, "hopkin": 189, "host": 2, "host1": 480, "host2": 480, "host_nam": [1, 2], "hostfil": 480, "hostnam": 480, "hot": 423, "hour": 5, "how": [2, 5, 6, 324, 326, 327, 330, 331, 332, 333, 334, 335, 340, 355, 356, 380, 400, 462, 479, 482, 487], "howev": [2, 112, 324, 341, 344, 463, 476, 479, 480, 483, 484], "hr": 343, "http": [344, 348, 350, 357, 382, 413, 435], "huber": 426, "huber_loss": 324, "human": [404, 405], "hundr": 8, "hurri": 5, "hutter": 459, "hyperbol": [0, 20, 22, 25, 107, 272, 290, 398, 447], "hz": 343, "i": [0, 1, 2, 3, 5, 6, 7, 8, 16, 18, 28, 37, 78, 83, 93, 99, 100, 101, 103, 104, 105, 108, 109, 110, 111, 112, 117, 118, 121, 122, 124, 125, 126, 127, 129, 136, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 161, 162, 163, 164, 168, 172, 173, 178, 179, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 194, 199, 203, 204, 208, 214, 217, 218, 232, 233, 236, 237, 238, 245, 246, 247, 257, 259, 262, 263, 264, 269, 273, 275, 280, 281, 286, 287, 288, 291, 294, 295, 299, 300, 301, 302, 303, 304, 307, 309, 310, 311, 312, 313, 317, 319, 324, 326, 327, 328, 330, 331, 332, 333, 334, 335, 336, 337, 338, 341, 343, 344, 348, 349, 350, 352, 355, 356, 362, 363, 369, 371, 372, 374, 375, 377, 378, 379, 381, 382, 383, 386, 391, 397, 399, 400, 404, 405, 411, 413, 421, 422, 424, 429, 430, 433, 434, 436, 441, 446, 452, 454, 456, 459, 461, 462, 463, 468, 470, 471, 476, 479, 480, 481, 482, 483, 484, 485, 486, 487, 488], "i386": 8, "i_n": 1, "i_nw": 1, "i_s": 1, "i_sw": 1, "i_t": 349, "iclr": [458, 459, 460], "id": [6, 8], "idea": [481, 483], "idempot": [363, 375], "ident": [0, 112, 140, 281, 324, 372], "identifi": [2, 310, 478], "idim": 6, "idiom": [6, 479], "idx": [38, 482], "ie": [375, 480], "ieee": 317, "ignor": [5, 38, 93, 94, 136, 456], "ih": 383, "ii": 1, "ij": 208, "imag": [0, 331, 332, 334, 335, 337, 338, 400], "imagenet": [404, 405], "imaginari": 170, "immedi": [5, 358], "implement": [0, 1, 4, 6, 145, 146, 189, 340, 362, 378, 386, 388, 391, 397, 399, 400, 446, 455, 456, 457, 458, 460, 461, 462, 474, 479, 481], "impli": 301, "implicit": [476, 479, 481], "implicitli": 483, "import": [2, 3, 4, 5, 6, 8, 112, 168, 189, 265, 299, 310, 311, 312, 313, 314, 324, 326, 327, 328, 348, 355, 356, 365, 400, 421, 423, 430, 449, 452, 454, 479, 480, 481, 482, 483, 484, 485], "improv": [1, 2, 3, 5, 421, 455, 456, 457, 458, 459, 460, 466, 479, 480], "in_ax": [303, 481], "in_channel": [330, 331, 332, 333, 334, 335], "in_dim": [324, 452], "in_proj": 452, "inci": 2, "includ": [1, 2, 108, 109, 110, 111, 143, 211, 212, 217, 350, 359, 371, 381, 424, 454, 479, 481, 482, 485, 486, 488], "include_dir": 2, "inclus": [0, 41, 42, 43, 44, 108, 109, 110, 111, 159], "incom": 2, "inconveni": 479, "incorpor": 484, "incorrect": 484, "increas": 218, "increment": 18, "incur": [5, 8], "incx": 2, "independ": [120, 337, 338], "index": [0, 1, 2, 7, 9, 28, 38, 138, 140, 165, 208, 233, 287, 288, 299, 315], "indic": [0, 2, 16, 26, 27, 28, 29, 38, 163, 164, 165, 172, 173, 174, 175, 176, 177, 178, 191, 236, 275, 287, 288, 299, 372, 374, 423, 430, 470, 482], "indices_or_sect": [71, 275], "indirectli": 484, "individu": [324, 337, 338], "ineffici": [482, 483], "inexact": [11, 178], "inf": [189, 225, 378], "infer": [7, 162, 194, 294, 480], "infin": [0, 174, 176, 177, 225, 355, 356, 460], "infinit": [16, 172, 173], "info": [5, 8], "inform": [3, 5, 6, 8, 131, 210, 263, 264, 317, 324, 328, 341, 378, 481, 487], "inherit": [6, 478], "inifn": 174, "init": [324, 379, 449, 454, 468, 469, 471, 472, 480], "init_fn": [401, 402, 403, 404, 405, 406, 407, 408, 449], "init_valu": 1, "initi": [1, 3, 4, 5, 123, 313, 324, 328, 344, 348, 350, 352, 379, 382, 401, 402, 403, 404, 405, 406, 407, 408, 452, 463, 468, 469, 471, 472, 479, 480, 483], "initializer_list": 0, "inject": 0, "inlin": 0, "inner": [0, 479], "inorm": 348, "inp": [1, 143], "inp_ndim": 1, "inp_shap": 1, "inp_strid": 1, "inplac": [2, 8], "input": [0, 1, 2, 4, 5, 12, 13, 14, 15, 16, 17, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 115, 117, 118, 121, 122, 127, 128, 129, 130, 131, 133, 134, 135, 137, 138, 139, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 163, 164, 165, 166, 167, 168, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 221, 222, 223, 224, 225, 226, 228, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 247, 250, 253, 254, 255, 256, 257, 258, 259, 260, 261, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 283, 284, 285, 287, 288, 289, 290, 291, 292, 293, 294, 295, 297, 298, 299, 300, 301, 303, 304, 306, 326, 327, 328, 330, 331, 332, 333, 334, 335, 337, 338, 340, 342, 343, 344, 348, 349, 350, 352, 355, 356, 378, 381, 382, 383, 386, 397, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 414, 421, 422, 424, 425, 426, 427, 429, 430, 432, 434, 446, 449, 479, 481, 482, 485, 486], "input_dil": [0, 101], "input_dim": [6, 324, 352, 381], "input_nam": [1, 143], "input_s": [343, 349, 383], "inputs1": 430, "inputs2": 430, "insert": [118, 138, 487], "insid": 479, "inspect": [3, 479, 485], "inspir": 7, "instabl": 434, "instal": 2, "instanc": [5, 38, 112, 237, 314, 324, 348, 358, 359, 360, 363, 365, 366, 367, 372, 375, 376, 377, 388, 452, 484], "instancenorm": 324, "instanti": [1, 2, 6, 483], "instantiate_axpbi": 2, "instead": [2, 8, 112, 324, 377, 391, 480, 481, 483], "int": [0, 1, 2, 5, 6, 9, 15, 17, 18, 26, 27, 28, 29, 33, 34, 35, 36, 41, 42, 43, 44, 45, 46, 49, 56, 57, 58, 59, 60, 63, 66, 68, 71, 74, 75, 76, 77, 78, 80, 83, 90, 91, 95, 98, 99, 100, 101, 102, 103, 104, 108, 109, 110, 111, 116, 117, 118, 125, 126, 127, 131, 138, 140, 141, 145, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 162, 164, 165, 169, 178, 185, 189, 193, 203, 205, 207, 210, 211, 212, 213, 216, 217, 218, 221, 223, 229, 232, 233, 235, 236, 237, 238, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 256, 257, 259, 260, 273, 274, 275, 278, 279, 280, 284, 285, 287, 288, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 303, 305, 307, 315, 324, 326, 327, 328, 330, 331, 332, 333, 334, 335, 340, 342, 343, 344, 348, 349, 350, 352, 355, 356, 378, 380, 381, 382, 383, 386, 391, 399, 414, 422, 423, 427, 432, 434, 452, 468, 470, 471, 472], "int16": 317, "int32": [0, 1, 11, 18, 38, 159, 178, 189, 248, 317, 400, 482, 485], "int64": [11, 317], "int8": [11, 317], "int_0": 134, "integ": [0, 11, 161, 163, 164, 178, 189, 210, 232, 237, 238, 240, 247, 248, 275, 287, 291, 303, 317, 340, 370, 470, 482], "integr": [18, 287, 483], "intend": [0, 479], "interact": 399, "interest": 487, "interfac": 2, "intermedi": 484, "intern": 328, "interpol": 400, "interv": [18, 193, 248, 252], "introduc": [0, 259], "intuit": 324, "invalid": [0, 83], "invers": [0, 19, 20, 21, 22, 23, 24, 25, 135, 150, 151, 152, 153, 154, 155, 184, 188, 192], "invert": 0, "involv": [454, 479], "iogpu": 218, "ip": 480, "is_avail": 123, "is_equival": 2, "is_floating_point": 2, "is_leaf": [310, 311, 312, 313], "is_leaf_fn": 362, "isclos": 0, "isfinit": 0, "ish": 5, "ishmael": 5, "isinf": 0, "isnan": 0, "isneginf": 0, "isposinf": 0, "issu": [480, 481, 484], "issubdtyp": [11, 317], "item": [0, 2, 4, 5, 6, 311, 454, 483, 484, 485], "iter": [4, 6, 191, 311, 312, 476, 479, 483], "iterm": 8, "itertool": [5, 311], "its": [0, 1, 2, 8, 184, 204, 233, 250, 296, 308, 314, 324, 381, 454, 458, 459, 460, 480, 484, 487], "itself": [2, 307, 463], "ix": 1, "ix_n": 1, "ix_nw": 1, "ix_s": 1, "ix_sw": 1, "iy_n": 1, "iy_nw": 1, "iy_s": 1, "iy_sw": 1, "j": [5, 8, 189, 337, 457, 458, 460], "j8": 2, "jacobian": [2, 179, 302, 485], "jain": 337, "jax": [7, 476], "jit": 143, "jmlr": 457, "jnp": 484, "john": 189, "join": 470, "join_schedul": 454, "jointli": 245, "just": [2, 6, 350, 479, 482], "jvp": [2, 112, 485], "k": [0, 5, 45, 90, 117, 140, 146, 163, 168, 293, 296, 297, 298, 326, 352, 355, 363], "k_h": [327, 356], "k_w": [327, 356], "kaim": 405, "keep": [2, 15, 17, 26, 27, 203, 205, 207, 221, 235, 280, 284, 300, 324, 362, 481, 483], "keepdim": [0, 15, 17, 26, 27, 33, 34, 35, 36, 56, 57, 58, 59, 63, 75, 76, 80, 189, 203, 205, 207, 221, 235, 273, 280, 284, 300], "kei": [1, 3, 5, 146, 210, 240, 241, 242, 244, 245, 246, 247, 248, 250, 251, 252, 310, 311, 362, 363, 375, 378, 463, 476, 478, 481], "kept": 218, "kernel": [2, 7, 8, 98, 99, 100, 101, 102, 103, 104, 143, 326, 327, 355, 356, 479, 482], "kernel_dil": [0, 101], "kernel_s": [326, 327, 330, 331, 332, 333, 334, 335, 355, 356], "key_cach": 5, "key_input_dim": 378, "key_proj": 5, "keyword": [165, 265, 266, 299, 311, 324, 476, 486, 488], "kind": 5, "kingma": [458, 460], "kl_div_loss": 324, "kname": 2, "know": [2, 5], "known": [389, 441], "kth": [0, 28, 233], "kullback": 427, "kw_onli": 2, "kwarg": [10, 120, 265, 266, 488], "l": [5, 6, 183, 184, 186, 187, 324, 326, 328, 330, 333, 343, 349, 355, 383, 433], "l1": [299, 426, 428, 429, 433], "l1_loss": 324, "l2": [426, 429, 467], "l2_loss": 324, "l_": [326, 355, 426], "la": 189, "label": [3, 4, 423, 430], "label_smooth": 423, "lack": 482, "lambd": [345, 395, 415, 445], "lambda": [311, 312, 313, 324, 345, 358, 363, 370, 395, 415, 439, 445, 455, 456, 457, 458, 459, 460, 461, 466, 467, 479, 480, 481], "languag": [1, 2], "larg": [5, 324, 378, 429, 479, 480, 483], "larger": [1, 145, 218, 386, 461], "largest": [189, 225, 293], "lasso": 299, "last": [0, 1, 5, 29, 78, 142, 144, 149, 152, 154, 155, 157, 158, 159, 163, 164, 171, 183, 184, 186, 187, 188, 190, 191, 192, 204, 213, 241, 274, 291, 301, 330, 331, 332, 333, 334, 335, 337, 338, 344, 400, 484], "latenc": 480, "later": [3, 8, 454], "launch": [1, 2, 123, 480, 482], "layer": [7, 142, 307, 324, 326, 327, 337, 338, 343, 344, 349, 350, 352, 355, 356, 372, 377, 380, 381, 383, 388, 399, 448, 452], "layer_s": 6, "layernorm": 324, "layout": 1, "lazi": [7, 452, 485], "lazili": [5, 324], "lceil": 90, "ld": [343, 349, 383], "ldot": [326, 327, 355, 356], "lead": [0, 18, 83, 479], "leaf": [94, 307, 310, 311, 312, 313, 362], "leaf_modul": 324, "leaki": [351, 418], "leaky_relu": 324, "leakyrelu": 324, "learn": [4, 6, 7, 328, 344, 348, 350, 379, 382, 454, 455, 456, 457, 458, 459, 460, 461, 466, 467], "learnabl": [330, 331, 332, 333, 334, 335, 388], "learning_r": [6, 454, 455, 456, 457, 458, 459, 460, 461, 463, 466, 467, 468, 469, 470, 471, 472, 479], "least": [5, 84, 85, 86, 93, 183, 184, 186, 187, 188, 190, 191, 192, 237], "leav": [2, 136, 311, 312, 313], "lectur": 466, "lecun": 337, "left": [0, 5, 145, 180, 189, 237, 259, 326, 327, 341, 355, 356, 386, 400, 412, 413, 424, 426, 434], "left_shift": 0, "leibler": 427, "len": [5, 149, 152, 155, 158, 168, 470], "length": [5, 278, 328, 330, 333, 343, 349, 383, 470], "leq": [426, 439], "less": [0, 5, 28, 182, 218, 233, 386, 433], "less_equ": 0, "let": [1, 2, 4, 5, 184, 479, 481, 483, 484], "level": [0, 163, 164, 404, 405], "lfloor": [326, 327, 355, 356], "lh": [343, 349, 383], "lhs_indic": [0, 163, 164], "lhs_mask": 90, "lib": 480, "libmlx": 8, "libmlx_ext": 2, "libmpi": 480, "librari": [2, 8, 319, 324], "like": [2, 5, 7, 126, 178, 230, 306, 338, 429, 463, 465, 479, 480, 481, 483, 484, 485, 487], "likelihood": [424, 432], "limit": [0, 2, 93, 216, 217, 218, 482], "linalg": 168, "line": [5, 480, 483, 484], "linear": [0, 2, 5, 6, 7, 307, 311, 324, 329, 339, 341, 342, 351, 365, 381, 383, 384, 385, 387, 389, 400, 409, 410, 411, 412, 413, 414, 418, 437, 438, 439, 441, 449, 452, 463, 471, 479], "linear1": 5, "linear2": 5, "linear3": 5, "linear_schedul": [454, 470], "linearli": 378, "link": [2, 8], "linspac": 0, "lion": 454, "list": [1, 5, 10, 15, 17, 30, 71, 78, 83, 84, 85, 86, 91, 94, 95, 101, 131, 136, 143, 148, 149, 151, 152, 154, 155, 157, 158, 162, 165, 179, 189, 203, 205, 207, 208, 221, 229, 232, 235, 240, 241, 242, 244, 245, 246, 248, 251, 252, 263, 273, 275, 279, 280, 284, 291, 292, 295, 299, 300, 302, 305, 310, 313, 314, 324, 363, 365, 366, 367, 368, 373, 375, 376, 377, 452, 454, 458, 459, 460, 461, 470, 478, 479, 480, 481, 483], "liter": [2, 232, 400, 404, 405, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434], "littl": 5, "liu": 5, "live": [7, 143, 487], "ll": [1, 4, 6, 426, 479, 481], "llama": 5, "llamaattent": 5, "llamaencoderlay": 5, "llm": 7, "load": [6, 7, 319, 365, 480], "load_weight": [324, 483], "loader": 6, "loader_path": 2, "loan": 189, "loc": [1, 244, 246], "local": [324, 337, 480], "locat": [0, 2, 83, 376, 377, 480, 487], "log": [0, 197, 199, 203, 353, 354, 419, 420, 421, 424, 427, 429, 432, 444], "log10": 0, "log1p": 0, "log2": 0, "log_cosh_loss": 324, "log_sigmoid": [324, 353], "log_softmax": [324, 354], "logaddexp": 0, "logarithm": [0, 195, 196, 197, 198], "logcosh": 429, "logic": [0, 2, 200, 201, 202], "logical_and": 0, "logical_not": 0, "logical_or": 0, "logist": [0, 4, 269, 413, 441], "logit": [5, 241, 421, 423, 479], "logsigmoid": 324, "logsoftmax": 324, "logsumexp": 0, "long": 5, "longer": [5, 105, 481], "look": [2, 5, 480], "lookup": 340, "loop": [5, 6, 479, 480, 481, 483], "loshchilov": 459, "loss": [4, 6, 299, 324, 454, 479, 480, 481, 483], "loss_and_grad": 324, "loss_and_grad_fn": [6, 454, 479, 481], "loss_fn": [4, 6, 454, 479, 481], "loss_grad_fn": 480, "lot": [480, 481], "low": [248, 252, 408, 449], "low_pad_s": 0, "lower": [183, 184, 186, 187, 192, 237, 248, 251, 252, 296, 408], "lr": [4, 461], "lr_schedul": [468, 469, 470, 472], "lstm": 324, "lto": 2, "lu": 5, "luckili": 483, "lvalu": 299, "m": [0, 2, 5, 8, 90, 140, 163, 168, 189, 296, 326, 327, 355, 356, 455, 479], "m1": [1, 5, 479, 481, 487], "m10": 317, "m7": 317, "m_": [458, 459, 460, 461], "m_t": [458, 459, 460, 461], "mac": 480, "machin": [5, 7, 8, 466, 480], "maco": [8, 218], "macosx": 8, "made": [5, 319], "mai": [2, 189, 307, 337, 480, 481, 482], "main": [7, 118, 140, 143, 294, 311, 312, 324, 480], "maintain": [337, 338, 461], "major": [0, 2], "make": [1, 2, 3, 5, 6, 8, 204, 227, 268, 324, 468, 469, 471, 472, 479, 483, 485, 487], "make_shar": 2, "malloc_or_wait": 2, "man": 5, "manag": [282, 476, 480, 487], "mani": [2, 83, 275, 330, 331, 332, 333, 334, 335, 340, 380, 479, 480, 483], "manual": 324, "map": [2, 6, 38, 194, 311, 340, 358], "map_fn": [358, 362], "map_torch_to_mlx": 5, "margin": [430, 434], "margin_ranking_loss": 324, "mask": [0, 5, 90, 146, 372, 378, 482], "mask_lh": [0, 90], "mask_n": 1, "mask_nw": 1, "mask_out": [0, 90], "mask_rh": [0, 90], "mask_s": 1, "mask_sw": 1, "matadata": 194, "match": [8, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 211, 365, 400, 423, 480, 482, 484], "materi": [5, 7], "math": [5, 434, 479], "mathbf": 184, "mathcal": 352, "mathemat": 189, "mathrm": [134, 269, 348], "matmul": [0, 163, 487], "matric": [189, 190, 191], "matrix": [0, 4, 14, 45, 90, 116, 117, 140, 141, 163, 164, 168, 169, 183, 184, 186, 187, 188, 189, 190, 191, 192, 204, 208, 237, 238, 245, 380, 381, 406, 449], "matter": [5, 324], "max": [0, 1, 2, 189, 206, 329, 355, 356, 379, 409, 416, 417, 422, 424, 425, 430, 434, 436, 438, 456, 460, 479, 481, 487], "max_": [355, 356], "max_buffer_s": 210, "max_freq": 391, "max_i": 237, "max_norm": 309, "max_recommended_working_set_s": [210, 218], "max_val": 416, "maximum": [0, 6, 26, 38, 93, 108, 213, 217, 309, 324, 351, 384, 391, 412, 413, 418, 437, 452, 483], "maxpool1d": 324, "maxpool2d": 324, "maxtotalthreadsperthreadgroup": 2, "mca": 480, "md": 189, "me": 5, "mean": [0, 1, 4, 5, 6, 144, 244, 245, 246, 299, 324, 328, 344, 363, 382, 407, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 479, 481, 484], "meant": 112, "measur": 487, "mechan": 399, "medic": 338, "meet": 8, "member": [2, 324, 368, 373], "memori": [0, 1, 2, 7, 83, 209, 211, 212, 213, 215, 216, 217, 218, 399, 452, 456, 479, 483, 484], "memory_order_relax": 1, "memory_s": [210, 218], "memoryview": [483, 484], "merg": 479, "meshgrid": 0, "metadata": [4, 194, 263, 264], "metal": [2, 7, 143], "metal_captur": 3, "metal_kernel": 1, "metal_path": 8, "metallib": [2, 8], "method": [2, 5, 9, 10, 30, 112, 120, 307, 315, 324, 325, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350, 351, 352, 353, 354, 355, 356, 357, 369, 378, 379, 380, 381, 382, 383, 384, 385, 386, 387, 388, 389, 390, 391, 392, 393, 394, 395, 396, 397, 398, 399, 400, 452, 455, 456, 457, 458, 459, 460, 461, 463, 466, 467, 474], "millisecond": [8, 479, 487], "min": [0, 2, 189, 222, 329, 379, 409, 416, 417, 436, 438], "min_freq": 391, "min_i": 237, "min_val": 416, "mind": [2, 5], "mine": 5, "minibatch": 6, "minim": 480, "minimum": [0, 27, 38, 93, 109, 391, 421, 422], "minsizerel": 8, "minu": 139, "minut": 5, "mish": 324, "miss": [365, 486], "mix": 482, "mkdir": [3, 8], "ml": 8, "mlp": [6, 324, 399, 454], "mlp_dim": [5, 399], "mlx": [1, 3, 4, 5, 6, 8, 319, 324, 449, 452, 454, 476, 478, 479, 480, 481, 482, 483, 484, 485, 486, 487], "mlx_build_benchmark": 8, "mlx_build_cpu": 8, "mlx_build_exampl": 8, "mlx_build_gguf": 8, "mlx_build_met": [2, 8], "mlx_build_metallib": 2, "mlx_build_python_bind": 8, "mlx_build_safetensor": 8, "mlx_build_test": 8, "mlx_disable_compil": [119, 132, 479], "mlx_ext": 2, "mlx_ext_metallib": 2, "mlx_include_dir": 2, "mlx_metal_debug": [3, 8], "mlx_metal_jit": 8, "mlx_sample_extens": 2, "mlx_trace": 3, "mnist": 6, "mode": [0, 1, 2, 105, 232, 361, 372, 374, 400, 404, 405, 480], "model": [4, 6, 7, 265, 307, 308, 311, 312, 324, 358, 361, 363, 365, 369, 372, 374, 375, 376, 378, 399, 449, 452, 454, 462, 463, 465, 479, 480, 483], "modest": 2, "modif": 484, "modifi": 484, "modul": [2, 5, 6, 307, 308, 388, 399, 449, 465, 478, 479, 483], "moment": [5, 456, 458, 459, 460], "momentum": [328, 461, 463, 467, 479], "monei": 5, "monoton": 435, "more": [1, 2, 3, 6, 10, 78, 118, 163, 183, 184, 186, 187, 188, 191, 192, 204, 216, 217, 263, 264, 317, 324, 328, 337, 386, 391, 399, 400, 402, 403, 404, 405, 421, 476, 479, 480, 481, 482, 485, 487], "most": [2, 241, 324, 465, 479, 480, 481, 482, 483], "move": [0, 2, 223, 487], "moveaxi": 0, "mpi": 319, "mpiexec": 480, "mpirun": 480, "mse": 299, "mse_loss": 324, "mtl": 2, "mtl_capture_en": 3, "mtlcommandbuff": 2, "mu": 467, "much": [1, 2, 5, 326, 327, 355, 356, 479, 483], "multi": [7, 146, 330, 331, 332, 333, 334, 335, 482, 484], "multidimension": 208, "multiheadattent": [5, 324], "multipl": [0, 1, 8, 14, 90, 142, 144, 163, 164, 204, 224, 237, 238, 378, 391, 469, 470, 472, 479, 483, 486], "multipli": [0, 2, 38, 164, 237, 238, 336, 391, 400], "murtadha": 5, "must": [0, 1, 2, 3, 8, 90, 93, 145, 162, 164, 186, 187, 189, 240, 241, 245, 248, 251, 252, 304, 400, 484], "mx": [1, 2, 3, 4, 5, 6, 38, 96, 97, 112, 123, 126, 143, 159, 178, 186, 187, 189, 190, 194, 247, 265, 299, 309, 324, 326, 327, 328, 339, 348, 351, 355, 356, 358, 365, 369, 384, 400, 401, 402, 403, 404, 405, 406, 407, 408, 410, 418, 421, 422, 423, 427, 430, 437, 447, 449, 452, 454, 476, 479, 480, 481, 482, 483, 484, 485, 486, 487, 488], "my": [5, 8], "my_devic": 488, "my_path": 265, "myexp": [1, 143], "myexp_strid": 1, "mymlp": 452, "n": [0, 1, 2, 5, 30, 90, 98, 99, 100, 101, 102, 103, 104, 140, 147, 149, 150, 152, 153, 156, 158, 168, 169, 245, 280, 296, 300, 326, 327, 328, 330, 331, 332, 333, 334, 335, 337, 338, 343, 349, 355, 356, 383, 400, 429, 434, 480], "n_i": [326, 327, 355, 356], "n_t": 343, "naiv": [2, 481], "naive_add": 481, "name": [1, 2, 143, 164, 194, 237, 238, 263, 264, 265, 266, 324, 344, 362, 365, 367, 480, 482, 486], "named_modul": 324, "nan": [0, 16, 82, 172, 173, 175, 225], "nan_to_num": 0, "nanobind": [2, 399], "nanobind_add_modul": 2, "nativ": 8, "natur": [0, 195, 197, 483], "nb": 2, "nb_domain": 2, "nb_func": 399, "nb_modul": 2, "nb_static": 2, "nbyte": 2, "nc": 328, "ndarrai": [30, 482, 483, 485], "ndhwc": [332, 335, 338], "ndim": [0, 1, 2, 159, 189, 191, 400], "ne": 1, "nearest": [1, 400], "necessari": 324, "necessarili": 293, "need": [1, 2, 5, 6, 7, 8, 82, 237, 324, 376, 377, 391, 399, 476, 480, 481, 483, 484, 485, 487], "neg": [0, 118, 159, 176, 225, 259, 294, 351, 355, 356, 378, 424, 432, 434, 482], "negat": [0, 226], "negative_slop": [351, 418], "neginf": [0, 225], "neighbor": 400, "neither": [165, 299], "nelem": 2, "nervou": 5, "nest": [78, 94, 313, 324, 452, 478, 481], "nesterov": 467, "network": [5, 7, 328, 337, 340, 402, 403, 449, 452, 466, 480], "neural": [5, 7, 340, 402, 403, 435, 449, 452, 466], "never": [5, 483], "new": [0, 2, 6, 91, 118, 223, 227, 257, 279, 295, 301, 311, 312, 370, 378, 452, 454, 465, 470, 479, 482, 483, 484], "new_tre": 312, "next": [2, 5, 6, 216], "nh": [343, 349, 383], "nhwc": [328, 331, 334], "nice": [481, 483], "nlc": [328, 330, 333], "nld": [343, 349, 383], "nlh": [343, 349, 383], "nll": [424, 432], "nll_loss": 324, "nn": [2, 5, 6, 265, 311, 324, 449, 452, 454, 463, 465, 479, 483], "nobodi": 5, "node": [94, 136, 303, 312, 313], "nois": 4, "noisi": 4, "nomins": 2, "non": [0, 1, 2, 8, 208, 373, 383, 435, 452], "none": [1, 2, 5, 9, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 115, 116, 117, 118, 119, 121, 122, 125, 126, 127, 128, 129, 130, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 215, 219, 220, 221, 222, 223, 224, 225, 226, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 264, 265, 267, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 303, 304, 305, 306, 307, 310, 311, 312, 313, 315, 326, 327, 341, 355, 356, 358, 362, 363, 370, 375, 378, 383, 391, 399, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 456, 474, 482], "nonlinear": [383, 479], "nonzero": 482, "noop": 375, "nor": [2, 165, 299], "norm": [5, 144, 309, 344, 434, 460, 461], "norm1": 5, "norm2": 5, "norm_first": 399, "normal": [1, 2, 4, 5, 142, 143, 144, 186, 245, 251, 324, 326, 327, 328, 344, 348, 350, 355, 356, 382, 399, 402, 404, 484, 487], "not_equ": 0, "notabl": [5, 7], "notat": [116, 141, 310, 367], "note": [0, 2, 5, 8, 16, 18, 83, 90, 94, 100, 103, 104, 112, 146, 154, 155, 164, 172, 189, 211, 237, 241, 301, 307, 324, 382, 400, 454, 484, 486], "noth": [5, 324, 483], "notic": [5, 481, 486], "now": [1, 2, 5, 8, 381, 479, 480, 484], "np": [1, 5, 6, 480, 484, 485], "npy": [194, 262, 486], "npz": [5, 194, 265, 266, 365, 369, 486], "nuclear": 189, "nullopt": 0, "num": [0, 5, 193, 250], "num_class": [6, 454], "num_decoder_lay": 399, "num_embed": [340, 380], "num_encoder_lay": 399, "num_epoch": [6, 454], "num_exampl": 4, "num_featur": [4, 328], "num_group": 344, "num_head": [5, 378, 399], "num_it": 4, "num_lay": [5, 6, 454], "num_param": 324, "num_paramet": 379, "num_sampl": 241, "num_split": 0, "number": [0, 2, 11, 18, 61, 70, 94, 99, 100, 101, 103, 104, 116, 140, 141, 164, 165, 169, 179, 193, 225, 232, 237, 238, 241, 244, 246, 250, 252, 256, 259, 260, 291, 292, 296, 299, 302, 303, 307, 324, 328, 330, 331, 332, 333, 334, 335, 337, 338, 344, 348, 378, 379, 399, 400, 402, 403, 404, 405, 468, 470, 471, 476, 479, 481, 488], "number_of_el": 0, "numer": [5, 142, 144, 189, 199, 203, 273, 328, 344, 348, 350, 382, 421, 422, 424, 434, 455, 456, 457, 458, 459, 460, 466, 479, 483], "numpi": [2, 5, 6, 7, 13, 16, 18, 87, 88, 89, 91, 128, 129, 133, 166, 167, 172, 180, 181, 182, 199, 204, 206, 222, 224, 228, 234, 255, 258, 283, 483, 485, 486], "nw": 1, "nwhc": 337, "o": [2, 8, 146, 349], "o_t": 349, "obj": 263, "object": [3, 10, 30, 50, 78, 94, 143, 178, 265, 303, 310, 311, 312, 313, 317, 337, 399, 478], "observ": 5, "occupi": [116, 141, 164, 237, 238], "occur": 484, "odim": 6, "odot": [343, 349], "off": [5, 8, 483], "offer": 429, "offset": [0, 1, 2, 5, 46, 83, 118, 142, 145, 294], "often": 338, "ok": [365, 481], "okai": [479, 483], "old": 5, "omit": [458, 460, 480], "onc": [2, 8, 479], "one": [0, 2, 5, 8, 38, 78, 84, 93, 99, 100, 101, 103, 104, 138, 140, 142, 144, 145, 189, 197, 204, 238, 241, 278, 283, 317, 375, 400, 423, 480, 487], "ones": [0, 2, 5, 230, 265, 296, 376, 377, 454, 480, 482], "ones_lik": 0, "onli": [1, 2, 5, 7, 8, 82, 90, 99, 100, 101, 103, 104, 186, 187, 189, 218, 237, 245, 301, 324, 362, 363, 365, 370, 372, 375, 376, 377, 452, 479, 480, 481, 486, 487], "onlin": 457, "op": [1, 2, 231, 301, 363, 483], "open": [3, 8, 18, 248, 252], "openmpi": 480, "oper": [3, 5, 7, 9, 37, 84, 85, 86, 101, 146, 163, 164, 234, 236, 273, 281, 288, 315, 324, 399, 461, 479, 480, 481, 482, 483, 484, 485, 487, 488], "operand": [130, 131, 163], "opportun": 479, "opt": [462, 480], "optim": [3, 4, 6, 7, 376, 479, 480, 481, 483], "option": [0, 3, 5, 14, 15, 17, 18, 26, 27, 28, 29, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 83, 84, 85, 86, 90, 94, 95, 98, 99, 100, 101, 102, 103, 104, 105, 108, 109, 110, 111, 112, 116, 117, 118, 121, 122, 123, 125, 126, 127, 140, 141, 142, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 162, 163, 164, 165, 169, 176, 177, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 203, 205, 207, 208, 217, 221, 225, 229, 232, 233, 235, 237, 238, 240, 241, 242, 244, 245, 246, 247, 248, 250, 251, 252, 256, 257, 259, 273, 274, 275, 278, 279, 280, 284, 286, 287, 291, 293, 294, 295, 296, 297, 298, 299, 300, 303, 305, 307, 310, 311, 312, 313, 326, 327, 328, 330, 331, 332, 333, 334, 335, 343, 349, 352, 355, 356, 358, 362, 363, 365, 370, 375, 378, 380, 381, 383, 386, 391, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 455, 456, 457, 458, 459, 460, 461, 463, 466, 467, 468, 476, 479, 486, 488], "ord": 189, "order": [0, 1, 28, 83, 101, 131, 186, 187, 189, 233, 237, 293, 324, 344, 376, 388, 463, 479, 481], "ordinari": 171, "org": [344, 348, 350, 357, 382, 413, 435], "origin": [5, 118, 309, 328, 371, 402, 403, 404, 405, 455, 456, 457, 458, 460, 461, 484], "orthonorm": 168, "ostream": 2, "ostringstream": 2, "other": [0, 2, 5, 7, 178, 189, 324, 364, 452, 461, 479, 480, 482, 483, 485], "other_input": 324, "otherwis": [18, 101, 123, 217, 247, 307, 310, 311, 312, 313, 363, 365, 375, 397, 399, 400, 415, 421, 426, 433, 445, 446, 483, 484], "our": [1, 2, 5, 6, 388, 455, 456, 457, 458, 460, 461, 480], "out": [0, 1, 2, 8, 90, 143, 326, 327, 337, 338, 355, 356, 372, 479, 480, 481, 482], "out_ax": [303, 481], "out_channel": [330, 331, 332, 333, 334, 335], "out_dim": [324, 452], "out_dtyp": 2, "out_idx": 2, "out_mask": 90, "out_proj": [5, 452], "out_ptr": 2, "out_shap": [1, 2], "outer": [0, 479, 483], "outlier": 429, "output": [0, 1, 2, 5, 8, 15, 16, 17, 18, 28, 83, 90, 91, 94, 96, 97, 108, 109, 110, 111, 112, 130, 140, 142, 143, 144, 145, 146, 153, 156, 157, 158, 162, 163, 165, 168, 169, 172, 189, 193, 203, 205, 207, 208, 221, 225, 229, 230, 233, 235, 236, 240, 241, 242, 244, 245, 246, 248, 251, 252, 265, 266, 273, 278, 280, 284, 288, 294, 296, 299, 300, 301, 302, 303, 304, 305, 306, 326, 327, 328, 330, 331, 332, 333, 334, 335, 348, 352, 355, 356, 378, 381, 397, 399, 400, 402, 403, 404, 405, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 446, 449, 479, 480, 481, 482, 483, 484, 485, 486, 487], "output_dim": [6, 324, 352, 381], "output_directori": 2, "output_dtyp": [1, 143], "output_fil": 5, "output_nam": [1, 143], "output_shap": [1, 143], "outsid": [143, 159], "over": [0, 2, 5, 6, 15, 17, 26, 27, 28, 29, 98, 99, 100, 101, 102, 103, 104, 108, 109, 110, 111, 149, 152, 155, 158, 171, 189, 191, 193, 203, 205, 207, 221, 233, 235, 261, 273, 274, 280, 284, 291, 293, 300, 328, 330, 331, 332, 333, 334, 335, 344, 350, 382, 423, 468, 471, 480, 481], "overal": 2, "overhead": [479, 483, 487], "overlap": 1, "overload": 18, "overrid": [2, 132], "overview": 3, "overwrit": 5, "own": [8, 484], "owndata": 484, "p": [8, 240, 324, 336, 337, 338, 434, 458, 460], "pack": [164, 237, 238], "packag": [2, 4, 6, 8, 319, 449, 480], "package_data": 2, "pad": [0, 1, 98, 99, 100, 101, 102, 103, 104, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 326, 327, 330, 331, 332, 333, 334, 335, 355, 356], "pad_valu": 0, "pad_width": [0, 232], "padding_hi": 0, "padding_lo": 0, "page": 485, "pain": 5, "pair": [0, 2, 232, 365, 386], "pairwis": 434, "pan": 5, "paper": [328, 391, 455, 456, 457, 458, 460, 461], "parallel": [480, 487], "param": [299, 324, 449, 481], "paramet": [0, 1, 2, 4, 5, 6, 12, 13, 14, 15, 16, 17, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 37, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 115, 116, 117, 118, 121, 122, 123, 125, 126, 127, 128, 129, 130, 131, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 216, 217, 218, 219, 221, 222, 223, 224, 225, 226, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 307, 308, 309, 310, 311, 312, 313, 314, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 348, 349, 350, 351, 352, 355, 356, 358, 359, 362, 363, 365, 370, 371, 372, 375, 376, 377, 378, 379, 380, 381, 382, 383, 386, 388, 391, 395, 397, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 414, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 446, 448, 449, 452, 454, 455, 456, 457, 458, 459, 460, 461, 462, 463, 465, 466, 467, 468, 469, 470, 471, 472, 474, 479, 480, 481, 483], "parameter_scal": 456, "parametr": [379, 436], "pars": 5, "parse_arg": 5, "parser": 5, "part": [1, 2, 170, 253, 481, 482], "partial": [376, 377, 479, 483], "particip": [121, 122, 125, 126, 127], "particular": [237, 344], "particularli": 479, "partit": [0, 28], "pass": [1, 2, 5, 6, 65, 79, 231, 232, 299, 308, 310, 311, 312, 324, 363, 375, 376, 377, 388, 479, 480, 483], "password": 480, "path": [3, 8, 131, 219, 265, 266, 307, 312, 365, 480], "pattern": [324, 483], "peak": [213, 215], "penalti": 467, "pep": 484, "per": [5, 6, 116, 141, 164, 237, 238, 307, 328, 344, 348, 350, 382, 474, 479, 480, 483], "perceptron": 7, "perf_count": 479, "perfectli": 483, "perform": [0, 1, 2, 3, 5, 7, 14, 90, 101, 108, 109, 110, 111, 127, 130, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 163, 164, 168, 186, 187, 204, 238, 260, 273, 287, 324, 344, 399, 404, 405, 454, 479, 480, 482, 483, 487], "perhap": [2, 5], "perm": 6, "permtuat": 247, "permut": [0, 6], "persist": 8, "pg": 189, "phi": [341, 411], "physic": 480, "pi": [134, 341, 391, 412, 481], "pick": 2, "pip": [2, 8], "pipelin": 2, "pixel": 337, "place": [2, 5, 38, 259, 260, 307, 480, 483, 484], "placehold": 479, "plai": [2, 5], "plain": 388, "plan": [2, 479], "platform": 8, "plu": [0, 197], "point": [0, 2, 4, 5, 8, 83, 161, 238, 317], "pointer": 2, "pool": [326, 327, 355, 356, 487], "popul": 2, "portion": 336, "posinf": [0, 225], "posit": [0, 5, 28, 118, 145, 159, 165, 177, 183, 184, 223, 225, 233, 245, 259, 294, 299, 311, 324, 330, 331, 332, 333, 334, 335, 378, 386, 391, 424, 434], "possibl": [275, 340, 380, 479, 480, 482, 487], "possibli": [5, 14, 90, 163, 204, 309], "postur": 5, "potenti": 217, "power": [0, 481, 484], "practic": [2, 479], "pre": [8, 146, 421], "preced": 344, "precis": [0, 2, 5, 139, 146, 324, 341, 382, 421, 462, 479], "preclud": 324, "pred": [425, 429], "predic": [307, 370], "predict": [421, 424, 425, 426, 427, 428, 429, 431, 432, 433], "prefix": [303, 310], "prelu": 324, "prepar": [2, 5], "prepend": [3, 204], "preprint": [5, 455, 461], "preprocessor": 8, "present": 1, "preserv": [257, 481], "press": [5, 189], "pressur": 2, "pretti": [479, 483], "prevent": [281, 434, 484], "previou": [216, 217, 218], "primal": [1, 2, 112, 179, 302], "primit": 481, "print": [1, 2, 4, 5, 6, 8, 309, 310, 311, 312, 314, 324, 476, 479, 480, 481, 482, 483, 484, 485], "prior": [236, 287, 288], "priorit": 481, "privat": 2, "prng": [240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 476], "prob": 421, "probabl": [8, 248, 336, 337, 338, 381, 421, 423, 427, 487], "problem": [4, 6, 324], "process": [5, 101, 105, 120, 121, 122, 123, 125, 126, 127, 311, 312, 338, 340, 399, 478, 480], "processor": 8, "prod": [0, 1], "produc": [0, 2, 8, 94, 378, 449], "product": [0, 2, 14, 83, 110, 171, 179, 185, 204, 231, 235, 291, 302, 378, 485], "profil": 3, "program": [213, 480], "programmat": 377, "project": [3, 5, 378], "project_source_dir": 2, "promot": 2, "promote_typ": 2, "promoted_dtyp": 2, "prompt": 5, "propag": [481, 482], "properti": [31, 38, 47, 51, 61, 62, 68, 70, 371, 374, 464, 481], "proportion": 309, "protocol": 484, "provid": [0, 2, 5, 83, 116, 141, 165, 247, 259, 291, 299, 311, 313, 319, 324, 358, 363, 365, 375, 376, 377, 380, 381, 399, 400, 448, 452, 480, 486, 488], "pseudo": 476, "pth": 5, "public": [2, 324], "pun": 0, "pure": [1, 324, 454], "purpos": [1, 189], "purs": 5, "push": 2, "push_back": 2, "put": [0, 1, 6, 236, 479, 480], "put_along_axi": 0, "py": [2, 5, 8, 480], "pypi": 8, "python": [1, 3, 5, 50, 68, 78, 136, 310, 311, 312, 313, 314, 452, 462, 463, 465, 478, 480, 481, 484], "python_requir": 2, "pytorch": [5, 7, 341, 344, 481], "pytorch_compat": 344, "q": [146, 190], "qualifi": 480, "quantiz": [0, 116, 141, 164, 194, 238, 380, 381], "quantized_matmul": 0, "quantizedembed": 324, "quantizedlinear": 324, "quarter": 5, "queri": [5, 146, 218, 378], "query_input_dim": 378, "query_proj": 5, "question": [5, 483], "queue": 3, "quick": [2, 7], "quit": [481, 484], "quotient": [0, 128, 129, 161], "r": [2, 5, 190, 299, 337, 343], "r_t": 343, "race": 487, "radian": [0, 115], "rag": 5, "rain": 5, "rais": [0, 5, 189, 217, 234, 275, 365], "ram": 5, "random": [1, 2, 3, 4, 5, 6, 7, 143, 326, 327, 328, 348, 355, 356, 365, 372, 479, 481, 487, 488], "randomli": [4, 5, 247, 336, 337, 338], "rang": [0, 2, 3, 4, 5, 6, 8, 18, 159, 163, 193, 403, 405, 412, 413, 454, 468, 469, 470, 471, 472, 476, 479, 481, 483, 487], "rank": [0, 125, 126, 127, 430, 480], "rate": [4, 454, 455, 456, 457, 458, 459, 460, 461, 466, 467], "rather": [2, 481, 487], "ratio": [0, 24], "rceil": 90, "re": [6, 8, 449], "readabl": 3, "readi": 2, "real": [0, 153, 154, 155, 156, 157, 158, 183, 184, 186, 187], "realli": 350, "reason": [1, 5, 482], "reboot": 8, "receiv": [125, 126, 307, 470, 484], "reciproc": [0, 261], "reclaim": 216, "recommend": [8, 217, 461], "recompil": [94, 479], "record": [3, 213, 483], "recreat": [314, 454], "rectifi": [351, 384, 385, 404, 405, 418, 437, 438], "recurr": [343, 349, 383], "recurs": [324, 362, 363, 368, 373, 375, 452], "recv": 126, "redirect": 2, "reduc": [0, 1, 8, 15, 17, 26, 27, 122, 203, 205, 207, 221, 235, 280, 284, 300, 313, 328, 399, 429], "reduct": [15, 17, 122, 203, 205, 221, 235, 313, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 480], "redund": 481, "refer": [189, 348, 357, 371, 402, 403, 404, 405, 413, 435, 482], "reflect": [371, 479, 482, 484], "regard": 341, "regardless": [83, 146], "regist": [2, 6], "register_librari": 2, "regress": [7, 429], "regular": [38, 337, 435, 459, 479, 482], "regularli": 2, "reimplement": 2, "rel": [16, 172, 456, 479], "relative_step": 456, "relax": 217, "relev": 2, "reli": [1, 2], "relu": [324, 379, 399, 436, 449], "relu6": 324, "remain": [0, 5, 218, 299, 312, 336, 337, 338, 480], "remaind": [0, 129], "remov": [0, 118, 204, 241, 278, 423], "rep": [0, 292], "repeat": [0, 292], "repeatedli": 4, "repetit": 256, "replac": [0, 5, 225, 376, 377, 399, 433], "replai": 3, "repli": 5, "repo": [4, 6, 8, 479], "report": [211, 217], "repres": [2, 5, 120, 123, 164, 430, 434, 484], "represent": [5, 237, 301, 310, 314], "request": 2, "requir": [1, 2, 5, 324, 480, 483, 484], "requires_grad": 481, "rerun": [479, 483], "rescal": 309, "research": 7, "reset": 215, "reset_peak_memori": 213, "reshap": [0, 5, 189, 400, 482], "resid": 218, "resolv": 2, "resourc": 2, "respect": [2, 4, 6, 142, 144, 163, 164, 165, 237, 299, 311, 324, 328, 341, 344, 348, 350, 452, 481, 485], "respons": 2, "rest": [5, 145, 311, 312, 386], "restart": 8, "restor": 259, "result": [0, 5, 14, 18, 38, 78, 83, 94, 142, 144, 164, 189, 204, 238, 245, 256, 279, 311, 312, 313, 391, 421, 479, 481, 484], "resum": 5, "return": [0, 1, 2, 4, 5, 6, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 37, 50, 68, 78, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 115, 116, 117, 118, 121, 122, 123, 125, 126, 127, 128, 129, 130, 131, 133, 134, 135, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 212, 216, 217, 218, 221, 222, 223, 224, 225, 226, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 260, 261, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 307, 308, 309, 310, 311, 312, 313, 314, 324, 343, 349, 358, 359, 360, 362, 363, 364, 365, 366, 367, 368, 372, 373, 375, 376, 377, 383, 401, 402, 403, 404, 405, 406, 407, 408, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 449, 452, 462, 478, 479, 480, 481, 482, 483, 484, 486, 487], "return_metadata": 194, "revers": [0, 2, 41, 42, 43, 44, 83, 108, 109, 110, 111, 295, 391], "rf": 8, "rfft": 153, "rfft2": 154, "rfftn": 155, "rfloor": [326, 327, 355, 356], "rho": 455, "rhs_indic": [0, 163, 164], "rhs_mask": 90, "right": [0, 1, 2, 8, 237, 258, 259, 326, 327, 341, 355, 356, 400, 412, 413, 424, 426, 434], "right_shift": 0, "rm": [5, 8, 144, 456], "rmsnorm": [5, 324], "rmsprop": 454, "rnn": [324, 343], "roadcast": 248, "robust": 429, "roform": [5, 386], "roll": 0, "root": [0, 5, 144, 261, 276, 382], "rope": [5, 324], "rosetta": 8, "rotari": [5, 145, 386], "rotat": [145, 386], "round": [0, 237], "routin": 2, "row": [0, 1, 2, 83, 140, 143, 169, 237, 296], "row_contigu": 2, "rpath": 2, "rsqrt": 0, "rtol": [0, 16, 172], "rule": [2, 454], "run": [1, 2, 3, 5, 6, 7, 8, 9, 143, 231, 315, 328, 358, 455, 456, 458, 459, 460, 479, 480, 483, 487, 488], "runtim": [5, 123, 319, 479, 480], "runtime_error": 2, "safetensor": [8, 194, 264, 365, 369, 454, 483, 486], "sai": [2, 5, 449, 483], "said": 5, "sake": 481, "same": [0, 2, 5, 8, 16, 38, 82, 91, 94, 99, 100, 101, 103, 104, 105, 121, 142, 144, 153, 156, 157, 158, 164, 165, 172, 179, 232, 241, 259, 260, 301, 302, 304, 312, 324, 327, 328, 336, 344, 348, 356, 380, 401, 402, 403, 404, 405, 406, 407, 408, 423, 434, 452, 462, 476, 479, 480, 482, 487], "sampl": [2, 4, 5, 193, 240, 241, 242, 244, 245, 248, 251, 252, 402, 403, 404, 405, 407, 408, 424, 430, 434, 476, 479], "sat": 5, "save": [3, 5, 7, 194, 219, 237, 263, 264, 265, 266, 369, 483], "save_gguf": 486, "save_safetensor": [369, 454, 486], "save_weight": 324, "savez": [5, 369, 486], "savez_compress": 486, "saw": [5, 481], "scalar": [0, 2, 13, 14, 16, 30, 50, 78, 82, 87, 88, 89, 90, 91, 93, 128, 129, 133, 161, 162, 165, 166, 167, 168, 172, 180, 181, 182, 193, 199, 200, 201, 202, 204, 206, 222, 224, 225, 228, 232, 234, 240, 248, 251, 252, 255, 258, 263, 283, 299, 301, 304, 308, 434, 481, 483, 485], "scale": [0, 2, 5, 14, 116, 141, 142, 144, 145, 146, 164, 168, 237, 238, 244, 246, 309, 337, 338, 350, 378, 386, 387, 391, 400, 439, 456], "scale_arr": 2, "scale_factor": 400, "scale_paramet": 456, "scatter": 0, "scatter_add": 0, "scatter_max": 0, "scatter_min": 0, "scatter_prod": 0, "schedul": [2, 217, 454, 468, 469, 470, 471, 472, 474, 487], "schema": 3, "scipi": 168, "scope": 324, "score": [5, 146, 430], "sdk": 8, "se": 1, "second": [5, 8, 118, 178, 180, 200, 202, 204, 258, 285, 294, 299, 327, 356, 422, 430, 456, 458, 459, 460, 481, 487], "second_layer_a": 483, "second_layer_b": 483, "secret": 5, "section": [1, 5, 8, 275, 434, 479, 480, 481], "see": [1, 2, 5, 6, 8, 10, 11, 32, 33, 34, 35, 36, 39, 40, 41, 42, 43, 44, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 189, 216, 263, 264, 307, 317, 324, 328, 329, 337, 339, 341, 345, 346, 347, 353, 354, 361, 379, 380, 381, 384, 385, 386, 387, 389, 391, 392, 393, 394, 395, 396, 398, 400, 402, 403, 404, 405, 411, 412, 413, 439, 479, 480, 481, 482, 485, 487], "seed": 243, "seen": 484, "select": [0, 3, 8, 186, 187, 293, 304, 358, 362, 370], "self": [5, 6, 9, 30, 31, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 50, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 112, 315, 324, 435, 452], "selu": 324, "semant": [13, 87, 88, 89, 91, 128, 129, 133, 166, 167, 180, 181, 182, 199, 204, 206, 222, 224, 228, 234, 255, 258, 283, 487], "semi": [183, 184, 245], "send": 480, "sennrich": 5, "sensit": 429, "sentencepiec": 5, "separ": [5, 65, 79, 344, 430], "sequenc": [5, 15, 17, 33, 34, 56, 57, 58, 59, 63, 71, 74, 75, 76, 80, 83, 91, 101, 125, 138, 143, 148, 149, 151, 152, 154, 155, 157, 158, 162, 203, 205, 207, 221, 229, 235, 240, 241, 242, 244, 245, 246, 248, 251, 252, 257, 273, 275, 278, 280, 284, 291, 292, 295, 300, 305, 328, 330, 333, 343, 349, 383, 399, 476, 487], "sequenti": [324, 449], "seri": 8, "serial": 454, "set": [2, 5, 6, 8, 94, 112, 119, 121, 122, 123, 125, 126, 127, 132, 142, 145, 210, 216, 217, 218, 267, 268, 282, 341, 350, 352, 361, 363, 370, 371, 372, 375, 376, 381, 386, 397, 422, 434, 446, 452, 454, 456, 463, 476, 481, 483], "set_data": 2, "set_default_devic": 2, "set_dtyp": 324, "set_input_arrai": 2, "set_memory_limit": 216, "set_output_arrai": 2, "setbyt": 2, "setcomputepipelinest": 2, "setup": [2, 4, 6, 8, 479], "sever": [5, 8, 98, 99, 100, 101, 102, 103, 104, 265, 266, 479, 486], "sgd": [4, 6, 454, 461, 463, 468, 469, 472, 479], "shade": [1, 2], "shall": 5, "shape": [0, 2, 3, 5, 6, 65, 82, 83, 90, 91, 94, 98, 99, 100, 101, 102, 103, 104, 118, 121, 125, 126, 143, 146, 147, 150, 153, 156, 157, 158, 162, 163, 168, 179, 188, 192, 204, 229, 230, 240, 241, 242, 244, 245, 246, 248, 251, 252, 257, 259, 301, 302, 304, 305, 306, 324, 326, 327, 328, 330, 331, 332, 333, 334, 335, 337, 338, 343, 348, 349, 352, 355, 356, 365, 383, 401, 402, 403, 404, 405, 406, 407, 408, 423, 434, 454, 479, 481, 482, 485, 487], "shapeless": [0, 94], "share": [7, 116, 141, 164, 237, 238, 301, 480], "shazeer": 5, "shift": [0, 180, 258, 259, 328], "shop": 5, "should": [2, 4, 5, 6, 8, 83, 118, 121, 142, 143, 144, 146, 179, 209, 218, 219, 236, 237, 288, 294, 299, 302, 307, 310, 324, 330, 331, 332, 333, 334, 335, 337, 338, 372, 378, 388, 423, 425, 430, 452, 478, 479, 480, 481, 483, 484, 488], "show": [8, 317, 479], "shown": 2, "shuffl": 6, "side": [0, 232, 326, 327, 355, 356, 479], "sigma": [341, 342, 343, 349, 390, 402, 403, 404, 405, 413, 414, 419, 440, 441], "sigmoid": [0, 5, 324, 353, 389, 413, 419, 421, 441], "sign": [0, 16, 172, 317, 461], "signal": [105, 400], "signatur": [1, 143], "signedinteg": [11, 178], "signific": 237, "silent": [156, 157, 158], "silicon": [2, 5, 7, 8, 487], "silu": 324, "simd": 1, "simd_sum": 1, "simdgroup": 1, "simdgroup_s": 1, "similar": [5, 164, 178, 311, 376, 377, 378, 422, 484, 486], "similarli": [2, 8, 204, 481, 483], "simpl": [2, 5, 6, 324, 340, 448, 454, 479, 480, 481, 483], "simple_axpbi": 2, "simple_tim": 2, "simplest": [2, 324, 480], "simpli": [2, 5, 8, 339, 351, 384, 410, 418, 437, 447, 452, 479, 480, 481], "simplic": 0, "simultan": 1, "sin": [0, 112, 391, 481, 485], "sinc": [1, 2, 5, 6, 164, 213, 452, 461, 470, 484, 487], "sine": [0, 21, 22, 271, 272, 481], "sing": 189, "singer": 457, "singl": [2, 6, 136, 179, 194, 208, 232, 302, 327, 356, 479, 482, 486], "singleton": [0, 15, 17, 26, 27, 123, 203, 204, 205, 207, 221, 235, 280, 284, 300], "singular": [189, 191], "sinh": 0, "sinusoid": 391, "sinusoidalpositionalencod": 324, "size": [0, 1, 2, 5, 6, 51, 68, 90, 99, 100, 103, 104, 116, 138, 141, 142, 143, 144, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 162, 164, 168, 169, 178, 185, 189, 212, 217, 218, 237, 238, 241, 257, 275, 278, 301, 307, 324, 326, 327, 330, 331, 332, 333, 334, 335, 340, 348, 355, 356, 380, 381, 400, 456, 480, 483, 484], "size_in_megabyt": 218, "size_t": [0, 2], "sizeof": 2, "skip": [3, 83], "slice": [0, 482], "slice_s": 0, "slice_upd": 0, "slight": [5, 483], "slightli": [386, 487], "slope": 351, "slot": 480, "slow": 479, "slowli": 5, "small": [5, 139, 142, 144, 328, 344, 350, 382, 424, 429, 434, 479, 480, 487], "smaller": [0, 8, 233, 461, 479], "smallest": 189, "smile": 5, "smooth": [423, 433, 466], "smooth_l1_loss": 324, "sned": 127, "snippet": 480, "so": [1, 2, 5, 8, 165, 168, 299, 336, 400, 454, 479, 480, 483, 487], "softmax": [0, 5, 146, 324, 354, 420, 423], "softmin": 324, "softplu": [324, 357, 435], "softshrink": 324, "softsign": 324, "solv": 324, "some": [0, 2, 4, 5, 6, 363, 375, 454, 463, 479, 481, 483], "someon": 5, "someth": [4, 5, 482], "sonoma": 8, "soon": 5, "sort": [0, 28, 29, 233, 293], "sourc": [0, 1, 2, 3, 60, 125, 126, 143, 223, 295, 480], "space": [0, 2, 193, 421, 432], "spars": [0, 208], "spatial": [99, 100, 101, 103, 104, 326, 344, 355, 400], "speak": [5, 189], "special": 2, "specif": [1, 2, 8, 480, 481], "specifi": [0, 2, 18, 37, 99, 100, 101, 103, 104, 118, 154, 155, 162, 165, 185, 189, 193, 223, 229, 236, 241, 256, 285, 287, 288, 291, 294, 295, 299, 303, 305, 328, 397, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 446, 480, 481, 487], "speed": [1, 2], "spent": 5, "split": [0, 342, 344, 414], "splittabl": 476, "sqrt": [0, 5, 134, 146, 168, 328, 341, 344, 348, 350, 352, 382, 391, 402, 403, 404, 405, 412, 455, 457, 458, 459, 466, 479], "squar": [0, 4, 5, 144, 169, 188, 192, 261, 276, 299, 311, 324, 382, 431, 433, 455, 456, 458, 459, 460, 481, 484], "squeez": [0, 400, 479], "src": [0, 125, 126], "ssh": 480, "stabil": [142, 144, 328, 344, 348, 350, 382, 421, 422, 424, 455, 456, 457, 458, 459, 460, 466], "stabl": [199, 203, 273, 429], "stable_abi": 2, "stack": [0, 479], "standard": [0, 1, 6, 50, 78, 204, 242, 246, 280, 399, 402, 404, 407, 480, 485], "starmap": [5, 311], "start": [0, 1, 2, 4, 5, 7, 8, 18, 145, 193, 219, 275, 313, 479, 482, 487], "start_axi": [0, 49, 159], "start_captur": 3, "state": [5, 6, 324, 343, 349, 383, 454, 463, 476, 479], "static": 8, "static_cast": 2, "std": [0, 2, 407], "step": [0, 3, 5, 6, 18, 324, 343, 349, 383, 456, 463, 468, 470, 471, 472, 479, 480], "step_decai": 454, "step_siz": 472, "still": [5, 8, 189, 479, 483], "stochast": [457, 458, 460, 467, 483], "stood": 5, "stop": [0, 2, 5, 18, 193, 220, 281, 481, 482], "stop_captur": 3, "stop_gradi": [0, 481], "storag": 83, "store": 5, "str": [2, 105, 130, 131, 143, 165, 186, 187, 189, 194, 208, 210, 219, 262, 263, 264, 265, 266, 299, 310, 314, 358, 359, 362, 363, 365, 367, 369, 375, 400, 404, 405, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434], "straight": 5, "strang": 5, "stream": [2, 7, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 114, 115, 116, 117, 118, 121, 122, 125, 126, 127, 128, 129, 130, 133, 134, 135, 137, 138, 139, 140, 141, 142, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 300, 301, 304, 305, 306, 480, 487], "streamcontext": 282, "streamordevic": [0, 2], "street": 5, "strength": [461, 467], "strict": [123, 166, 181, 363, 365, 375], "strictli": [189, 218], "stride": [0, 2, 83, 98, 99, 100, 101, 102, 103, 104, 326, 327, 330, 331, 332, 333, 334, 335, 355, 356, 386, 482], "string": [0, 2, 131, 143, 210, 232, 484, 486], "structur": [2, 462, 481], "stub": 8, "style": [2, 13, 16, 87, 88, 89, 128, 129, 133, 166, 167, 172, 180, 181, 182, 199, 204, 206, 222, 224, 228, 234, 255, 258, 283], "su": 5, "sub": [0, 6, 118, 250, 294, 307], "subarrai": [118, 275], "subclass": 452, "subdtyp": 178, "subgradi": 457, "sublinear": 456, "submodul": [5, 6, 324, 359, 363, 364, 375, 377], "subscript": [130, 131], "subsect": 5, "subsequ": 454, "subset": [324, 362], "substanti": 8, "subtract": [0, 38], "subtyp": [178, 317], "sudo": [8, 218], "sum": [0, 2, 4, 13, 111, 122, 171, 189, 203, 273, 291, 294, 324, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 480, 482, 484], "sum_": [189, 326, 327, 429], "sum_i": 420, "sum_j": [442, 443], "summat": [130, 131], "super": [5, 6, 324, 452], "superset": [311, 462], "support": [1, 2, 5, 7, 8, 16, 90, 100, 103, 104, 146, 159, 168, 172, 183, 184, 186, 187, 188, 190, 191, 192, 194, 204, 237, 245, 480, 481, 482, 484, 486], "suppos": [481, 487], "sure": [2, 3, 5, 8, 324, 479], "surpass": [404, 405], "surpris": 5, "sw": 1, "swap": [0, 105, 217, 285, 377], "swapax": [0, 112], "swiglu": 5, "swish": [389, 441], "switch": 8, "symbol": 461, "symmetr": [99, 100, 103, 104, 183, 184, 186, 187], "symmetri": [186, 187], "synchron": [2, 479], "syntax": [38, 482], "synthet": 4, "sysctl": 218, "system": [5, 8, 210, 211, 212, 218], "t": [0, 1, 2, 5, 8, 134, 143, 146, 164, 183, 184, 238, 299, 324, 326, 343, 349, 355, 383, 455, 456, 457, 458, 459, 460, 461, 466, 467, 479, 481, 487], "tabl": [1, 189, 317, 340], "take": [0, 2, 5, 6, 87, 88, 89, 94, 163, 165, 179, 206, 222, 230, 238, 288, 299, 302, 303, 306, 312, 313, 378, 421, 476, 480, 481, 482, 486, 487, 488], "take_along_axi": [0, 482], "taken": [118, 287, 294], "talk": 480, "tan": 0, "tangent": [0, 2, 23, 24, 25, 112, 179, 289, 290, 398, 447], "tangent_i": 2, "tangent_x": 2, "tanh": [0, 324, 341, 343, 349, 357, 383, 412, 435], "target": [2, 299, 421, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 479], "target_include_directori": 2, "target_link_librari": 2, "target_link_opt": 2, "target_sourc": 2, "task": [217, 429], "tau": 467, "tcp": 480, "tell": [5, 479, 484], "temp": 5, "templat": [0, 1, 2, 143], "ten": [481, 483], "tend": 461, "tensor": [194, 291, 326, 327, 355, 356, 434, 484], "tensordot": 0, "term": [2, 424, 455, 456, 457, 458, 459, 460, 466], "termin": 8, "test": [6, 8, 480], "test_imag": 6, "test_label": 6, "text": [5, 326, 327, 341, 343, 349, 355, 356, 357, 383, 390, 397, 402, 403, 404, 405, 412, 415, 416, 417, 424, 425, 426, 429, 430, 433, 435, 436, 439, 440, 445, 446, 456, 461], "textrm": [237, 341, 342, 411, 414], "tf": 484, "tgp_size": 2, "th": [108, 109, 110, 111, 117, 140, 186, 470], "than": [1, 2, 5, 78, 105, 118, 129, 145, 163, 166, 167, 181, 182, 183, 184, 186, 187, 188, 191, 192, 204, 216, 218, 309, 311, 386, 397, 400, 430, 433, 446, 456, 461, 479, 481, 487], "thank": 483, "thei": [1, 2, 4, 5, 8, 16, 105, 164, 172, 388, 425, 452, 461, 478, 479, 480, 483, 485, 486, 487], "them": [0, 2, 5, 121, 324, 363, 375, 480, 487], "themselv": [2, 479], "thi": [0, 1, 2, 5, 6, 8, 15, 16, 17, 18, 26, 27, 28, 29, 83, 112, 132, 143, 163, 164, 168, 172, 179, 183, 184, 186, 187, 188, 189, 190, 191, 192, 199, 203, 204, 205, 207, 209, 211, 218, 221, 233, 235, 241, 268, 273, 274, 275, 280, 284, 287, 293, 300, 309, 312, 313, 324, 336, 337, 338, 342, 343, 349, 359, 360, 362, 363, 366, 367, 368, 373, 375, 376, 377, 378, 381, 383, 397, 402, 403, 404, 405, 412, 413, 414, 421, 429, 446, 452, 463, 478, 479, 480, 481, 483, 484, 486], "thing": [2, 5, 480], "third": 185, "thompson": 337, "those": [2, 5, 324], "though": [2, 5, 479, 483, 484], "thousand": 483, "thread": [1, 2], "thread_index_in_simdgroup": 1, "thread_position_in_grid": [1, 2, 143], "threadgroup": [1, 2, 143], "threads_per_simdgroup": 1, "three": [5, 86, 400], "threefri": 476, "threshold": [397, 426, 433, 446], "through": [1, 2, 281, 399, 461, 479, 481, 484], "throw": [2, 94, 123], "thu": [5, 324], "thumb": 454, "tic": 479, "tieleman": 466, "tile": [0, 146], "time": [2, 5, 8, 217, 292, 324, 326, 327, 343, 349, 355, 356, 383, 479, 481, 483, 487], "timeit": [479, 481], "titl": 2, "tmp": [1, 143], "to_quant": 307, "to_stream": 2, "toc": 479, "togeth": [0, 1, 2, 6, 237, 311, 312, 480], "tok_embed": 5, "token": [5, 340, 380], "told": 5, "toler": [0, 16, 172], "too": [178, 479, 483], "took": 5, "tool": 8, "top": [2, 293, 352, 400], "topk": 0, "torch": [5, 484], "torch_weight": 5, "total": [218, 481], "total_norm": 309, "tpi": 479, "trace": [0, 3, 479], "trace_fil": 3, "tracer": 376, "track": [2, 324, 328], "track_running_stat": 328, "trade": 483, "tradit": [5, 145, 337, 338, 386], "train": [5, 6, 324, 328, 336, 337, 338, 361, 363, 375, 402, 403], "train_imag": [6, 454], "train_label": [6, 454], "trainabl": [6, 308, 324, 452], "trainable_paramet": [324, 362, 463], "transform": [1, 5, 7, 112, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 168, 308, 324, 328, 344, 350, 352, 362, 363, 375, 381, 386, 482], "transformerencod": 265, "transit": 470, "translat": [142, 350], "transpos": [0, 5, 31, 102, 103, 104, 164, 238, 333, 334, 335], "treat": [0, 2, 154, 155, 157, 158, 287, 400, 479], "tree": [7, 94, 136, 165, 299, 303, 310, 311, 312, 313, 314, 462, 463, 465, 474, 481], "tree_flatten": [265, 311, 314, 324, 454], "tree_map": [312, 324, 480], "tree_unflatten": [5, 454], "trembl": 5, "tri": 0, "triangl": [186, 187, 296], "triangular": [183, 184, 192], "tril": 0, "trilinear": 400, "triplet": 434, "triplet_loss": 324, "triu": 0, "true": [0, 1, 2, 4, 5, 16, 41, 42, 43, 44, 82, 94, 108, 109, 110, 111, 143, 145, 164, 172, 178, 183, 184, 189, 194, 208, 217, 238, 273, 304, 307, 310, 311, 312, 313, 317, 324, 328, 330, 331, 332, 333, 334, 335, 343, 344, 348, 349, 350, 352, 362, 363, 365, 372, 375, 381, 383, 386, 391, 399, 400, 421, 429, 456], "truncat": [147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 251], "truth": [4, 423, 433], "try": [2, 8], "tupl": [0, 30, 65, 68, 79, 95, 99, 100, 101, 103, 104, 125, 129, 131, 136, 138, 179, 186, 189, 190, 191, 232, 237, 257, 259, 278, 299, 302, 310, 311, 312, 313, 314, 326, 327, 331, 332, 334, 335, 355, 356, 365, 367, 388, 400, 456, 458, 459, 460, 461, 478, 481], "tutori": 2, "twice": 487, "two": [0, 2, 13, 14, 16, 24, 82, 85, 87, 88, 89, 90, 118, 128, 133, 148, 151, 157, 163, 164, 166, 167, 172, 181, 182, 183, 184, 185, 186, 187, 188, 190, 191, 192, 199, 204, 206, 222, 224, 228, 231, 285, 313, 327, 342, 349, 356, 414, 422, 479, 480, 481, 482, 487], "txt": 2, "type": [0, 1, 2, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 37, 68, 78, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 115, 116, 117, 118, 121, 122, 123, 125, 126, 127, 128, 129, 130, 131, 133, 134, 135, 137, 138, 139, 140, 141, 142, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 210, 216, 217, 218, 221, 222, 223, 224, 225, 226, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 260, 261, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 283, 284, 285, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 309, 310, 313, 324, 370, 399, 401, 402, 403, 404, 405, 406, 407, 408, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 479, 482], "type_nam": 2, "type_to_nam": 2, "typenam": [0, 1, 2], "typic": [0, 146, 340, 454, 479, 483], "u": [1, 2, 183, 186, 187, 191, 352, 377, 474, 483], "u_": 455, "u_t": 455, "uint": [1, 2, 143], "uint16": [11, 317], "uint3": 1, "uint32": [11, 26, 27, 28, 29, 241, 317], "uint64": [11, 317], "uint8": [11, 317], "ultra": 5, "unabl": 8, "unam": 8, "unari": 479, "unchang": [145, 281, 386], "uncheck": 8, "uncompress": 265, "undefin": [0, 28, 112, 183, 184, 233, 245, 482], "under": [2, 189], "underli": [2, 301], "understand": [5, 402, 403], "unexpect": [2, 18], "unfreez": [324, 363], "unfrozen": 375, "unifi": 7, "uniform": [3, 324, 352, 365, 403, 405, 449, 476, 479, 481, 487], "uniformli": 252, "unintend": 0, "union": [18, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 84, 85, 86, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 176, 177, 178, 186, 187, 210, 263, 282], "uniqu": [2, 476], "unique_ptr": 2, "unit": [329, 339, 341, 342, 343, 351, 384, 385, 387, 389, 402, 403, 404, 405, 409, 410, 411, 412, 413, 414, 418, 437, 438, 439, 441], "unittest": 8, "univers": 189, "unless": [5, 16, 172, 189, 452], "unlik": [5, 16, 172, 337, 338, 371], "unnecessari": [2, 5], "unnorm": [241, 421, 423], "unscal": 456, "unsign": [164, 237, 238, 317], "unsignedinteg": 11, "unspecifi": [15, 17, 18, 26, 27, 28, 29, 95, 108, 109, 110, 111, 162, 203, 205, 207, 221, 229, 233, 235, 256, 273, 274, 280, 284, 287, 293, 294, 300, 305, 488], "unsqueez": 5, "unsupport": 194, "until": [2, 483, 485], "unus": 2, "up": [1, 2, 5, 112, 479], "upcast": 2, "updat": [0, 1, 2, 4, 5, 6, 38, 94, 307, 311, 313, 328, 358, 359, 365, 370, 371, 372, 377, 454, 456, 459, 461, 462, 463, 467, 468, 469, 470, 471, 472, 479, 480, 483], "update_modul": 324, "uplo": [186, 187], "upon": [5, 311, 312], "upper": [183, 184, 186, 187, 192, 237, 248, 251, 252, 408], "upsampl": 324, "us": [0, 3, 4, 5, 6, 7, 8, 18, 38, 83, 112, 116, 119, 121, 122, 125, 126, 127, 129, 141, 143, 145, 159, 164, 180, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 204, 211, 212, 213, 216, 218, 237, 238, 256, 257, 258, 259, 286, 310, 313, 317, 319, 324, 327, 337, 340, 341, 343, 349, 352, 356, 358, 362, 369, 376, 378, 380, 381, 383, 386, 391, 399, 400, 404, 405, 412, 413, 422, 449, 452, 454, 455, 456, 458, 459, 460, 461, 462, 463, 476, 478, 479, 480, 481, 482, 485, 487], "usag": [112, 399, 479], "user": [2, 5, 324], "usual": [340, 380, 478, 483], "util": [1, 2, 5, 7, 8, 265, 324, 454, 480], "v": [5, 105, 146, 186, 324, 363, 484], "v_": [455, 457, 458, 459, 460, 466, 467], "v_t": [455, 457, 458, 459, 460, 466, 467], "val": [0, 30, 162], "valid": [6, 105, 159, 303, 310, 363, 375, 478], "valid_parameter_filt": 358, "valu": [0, 1, 4, 5, 11, 12, 16, 18, 26, 27, 50, 78, 82, 93, 140, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 162, 172, 185, 189, 191, 193, 210, 218, 225, 232, 236, 240, 241, 242, 244, 245, 246, 248, 251, 252, 259, 263, 287, 288, 299, 303, 308, 310, 311, 312, 313, 317, 327, 329, 336, 337, 338, 339, 345, 348, 352, 356, 362, 378, 379, 395, 397, 399, 401, 421, 422, 423, 424, 425, 426, 428, 429, 430, 431, 432, 433, 446, 452, 456, 459, 468, 469, 471, 472, 481], "value_and_grad": [6, 112, 324, 376, 452, 454, 465, 479, 481, 484, 485], "value_and_grad_fn": 483, "value_cach": 5, "value_dim": 378, "value_input_dim": 378, "value_output_dim": 378, "value_proj": 5, "valueerror": [189, 365, 481], "values_hat": 5, "van": 189, "var": [0, 328, 344, 348, 350, 424], "variabl": [8, 94, 119, 132, 165, 179, 299, 302, 303, 480], "varianc": [0, 280, 300, 328, 344, 424], "variant": [5, 433, 460], "variou": 189, "vector": [0, 2, 4, 7, 171, 179, 189, 287, 302, 303, 340, 423, 485], "verbos": [1, 143], "veri": [5, 378, 480, 483, 487], "verifi": [4, 8], "versa": 259, "version": [2, 8, 116, 141, 199, 203, 237, 273, 303, 476, 481, 482], "versu": 479, "via": [8, 112, 462, 465, 480, 483, 484], "vice": 259, "video": 338, "view": [0, 3, 83, 484], "virtual": 2, "vjp": [2, 112, 485], "vmap": [2, 112, 481, 483, 485], "vmap_add": 481, "vocab_s": 5, "vocabulari": [340, 380], "void": [1, 2], "vt": 191, "w": [0, 1, 4, 99, 100, 103, 104, 116, 141, 164, 186, 237, 238, 299, 312, 327, 328, 331, 332, 334, 335, 337, 338, 352, 356, 454, 467, 481], "w1": [5, 309], "w2": [5, 309], "w3": 5, "w_": [327, 343, 349, 356, 383, 455, 456, 457, 458, 459, 460, 461, 466, 467], "w_1": 237, "w_g": 237, "w_i": [116, 141, 237], "w_in": 1, "w_q": 237, "w_star": 4, "w_stride": 1, "w_t": [455, 457, 458, 459, 460, 461, 466, 467], "wa": [5, 83, 125, 126, 480, 483], "wai": [2, 5, 8, 324, 400, 479, 480, 481, 482], "wait": [2, 5, 217], "walk": 5, "walkthrough": 2, "walsh": 168, "want": [1, 5, 480, 481, 487], "warm": [2, 479], "warmup": [470, 471], "warmup_init": 456, "watch": [5, 479], "wd": 461, "we": [0, 1, 2, 4, 5, 6, 116, 125, 126, 141, 164, 237, 238, 324, 340, 380, 388, 459, 461, 476, 478, 479, 480, 481, 483, 487], "weight": [0, 4, 98, 99, 100, 101, 102, 103, 104, 142, 144, 311, 324, 365, 369, 380, 381, 421, 423, 452, 456, 459, 461, 463, 467, 481, 483], "weight_decai": [456, 459, 461, 467], "weight_fil": 5, "weights_fp16": 483, "well": [5, 324, 363, 375, 378, 483], "wen": 5, "went": 5, "were": [5, 487], "wet": 5, "what": [2, 5, 311], "whatsoev": 5, "whc": 337, "when": [0, 1, 2, 5, 7, 8, 94, 101, 112, 127, 183, 184, 186, 187, 188, 189, 191, 192, 194, 330, 331, 332, 333, 334, 335, 400, 404, 405, 421, 427, 433, 452, 454, 470, 476, 479, 480, 487], "where": [0, 6, 140, 172, 184, 237, 299, 303, 326, 327, 328, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 344, 348, 349, 350, 352, 355, 356, 362, 379, 382, 383, 397, 404, 405, 410, 411, 413, 424, 430, 436, 439, 441, 446, 463, 480, 481, 482], "wherea": 481, "whether": [143, 164, 186, 187, 192, 238, 343, 349, 362, 378, 383, 421, 424, 430], "which": [0, 1, 2, 5, 6, 7, 8, 18, 37, 83, 94, 101, 118, 121, 122, 125, 126, 127, 136, 145, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 165, 173, 174, 175, 176, 177, 179, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 194, 208, 219, 237, 241, 242, 256, 257, 259, 262, 263, 264, 265, 266, 278, 279, 287, 294, 299, 302, 303, 307, 327, 337, 338, 341, 356, 358, 362, 386, 421, 423, 426, 430, 433, 449, 462, 463, 476, 479, 480, 481, 482, 483, 487, 488], "while": [2, 3, 5, 8, 257, 386, 483, 484], "whistl": 2, "who": 5, "whose": [140, 307, 308], "why": 5, "wide": 483, "width": [327, 328, 331, 332, 334, 335, 337, 338, 356, 380, 381], "window": [8, 326, 327, 355, 356], "wipe": 8, "wire": 218, "wired_limit_mb": 218, "wise": [0, 2, 12, 13, 19, 20, 21, 22, 23, 24, 25, 87, 88, 89, 92, 106, 107, 128, 129, 133, 134, 135, 137, 139, 160, 161, 166, 167, 172, 180, 181, 182, 195, 196, 197, 198, 199, 200, 201, 202, 206, 222, 224, 226, 228, 234, 254, 255, 258, 261, 269, 270, 271, 272, 276, 277, 283, 289, 290, 329, 337, 338, 347, 357, 379, 390, 409, 416, 417, 419, 420, 435, 436, 438, 441, 442, 443, 444, 479], "wish": 8, "with_logit": 421, "within": [0, 3, 28, 172], "without": [1, 5, 7, 281, 378, 448, 478, 479, 480, 483, 484, 487], "wk": 5, "wl": 2, "wo": 5, "word": 0, "work": [2, 3, 5, 217, 479, 480, 481, 482, 483], "workhors": 324, "world": [314, 480], "worri": [1, 483], "would": [2, 5, 400, 480, 482, 483, 484, 487], "wq": 5, "wrap": [112, 324], "write": [0, 1, 2, 5, 324, 484], "written": 2, "wrt": 308, "wv": 5, "x": [0, 1, 2, 4, 5, 6, 38, 90, 112, 121, 122, 126, 127, 134, 139, 142, 143, 144, 164, 168, 169, 189, 238, 242, 247, 260, 265, 269, 297, 298, 304, 311, 313, 324, 326, 327, 328, 329, 339, 341, 342, 344, 348, 350, 351, 352, 355, 356, 357, 358, 379, 382, 384, 390, 391, 397, 400, 409, 410, 411, 412, 413, 414, 415, 416, 417, 418, 419, 420, 433, 435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445, 446, 447, 452, 454, 461, 479, 480, 481, 482, 483, 484, 485, 487], "x1": 422, "x2": 422, "x86_64": 8, "x_1": [422, 430], "x_2": [422, 430], "x_cast": 2, "x_grad": 1, "x_i": [420, 442, 443], "x_j": [442, 443], "x_offset": 2, "x_ptr": 2, "x_shape": 1, "x_stride": 2, "x_t": [343, 349, 383], "x_view": 484, "xcode": 8, "xcodeproj": 3, "xcrun": 8, "xf": 349, "xg": 349, "xi": 349, "xn": 343, "xo": 349, "xor": 89, "xr": 343, "xy": [0, 208], "xz": 343, "x\u00b2": 484, "y": [0, 2, 4, 5, 6, 38, 112, 168, 304, 324, 328, 337, 344, 348, 350, 352, 382, 425, 430, 433, 454, 457, 479, 480, 481, 483, 484], "y_": [425, 429], "y_cast": 2, "y_hat": 324, "y_offset": 2, "y_ptr": 2, "y_stride": 2, "ye": 5, "year": 5, "yet": [5, 189, 324, 452, 463, 481, 482, 483, 485], "yield": [5, 6, 476], "you": [2, 3, 5, 6, 7, 8, 218, 324, 391, 399, 449, 476, 479, 480, 481, 482, 484, 486, 487], "your": [2, 5, 8, 452, 481, 483], "z": [2, 343, 479, 483], "z_t": 343, "zeiler": 455, "zero": [0, 140, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 185, 208, 215, 296, 297, 298, 306, 324, 326, 327, 336, 337, 338, 365, 401, 402, 403, 404, 405, 406, 407, 408, 449, 454, 456, 482], "zero_grad": 481, "zeros_lik": 0, "zhang": 5, "zip": [5, 6], "zip_saf": 2}, "titles": ["Operations", "Custom Metal Kernels", "Custom Extensions in MLX", "Metal Debugger", "Linear Regression", "LLM inference", "Multi-Layer Perceptron", "MLX", "Build and Install", "mlx.core.Device", "mlx.core.Dtype", "mlx.core.DtypeCategory", "mlx.core.abs", "mlx.core.add", "mlx.core.addmm", "mlx.core.all", "mlx.core.allclose", "mlx.core.any", "mlx.core.arange", "mlx.core.arccos", "mlx.core.arccosh", "mlx.core.arcsin", "mlx.core.arcsinh", "mlx.core.arctan", "mlx.core.arctan2", "mlx.core.arctanh", "mlx.core.argmax", "mlx.core.argmin", "mlx.core.argpartition", "mlx.core.argsort", "mlx.core.array", "mlx.core.array.T", "mlx.core.array.abs", "mlx.core.array.all", "mlx.core.array.any", "mlx.core.array.argmax", "mlx.core.array.argmin", "mlx.core.array.astype", "mlx.core.array.at", "mlx.core.array.conj", "mlx.core.array.cos", "mlx.core.array.cummax", "mlx.core.array.cummin", "mlx.core.array.cumprod", "mlx.core.array.cumsum", "mlx.core.array.diag", "mlx.core.array.diagonal", "mlx.core.array.dtype", "mlx.core.array.exp", "mlx.core.array.flatten", "mlx.core.array.item", "mlx.core.array.itemsize", "mlx.core.array.log", "mlx.core.array.log10", "mlx.core.array.log1p", "mlx.core.array.log2", "mlx.core.array.logsumexp", "mlx.core.array.max", "mlx.core.array.mean", "mlx.core.array.min", "mlx.core.array.moveaxis", "mlx.core.array.nbytes", "mlx.core.array.ndim", "mlx.core.array.prod", "mlx.core.array.reciprocal", "mlx.core.array.reshape", "mlx.core.array.round", "mlx.core.array.rsqrt", "mlx.core.array.shape", "mlx.core.array.sin", "mlx.core.array.size", "mlx.core.array.split", "mlx.core.array.sqrt", "mlx.core.array.square", "mlx.core.array.squeeze", "mlx.core.array.std", "mlx.core.array.sum", "mlx.core.array.swapaxes", "mlx.core.array.tolist", "mlx.core.array.transpose", "mlx.core.array.var", "mlx.core.array.view", "mlx.core.array_equal", "mlx.core.as_strided", "mlx.core.atleast_1d", "mlx.core.atleast_2d", "mlx.core.atleast_3d", "mlx.core.bitwise_and", "mlx.core.bitwise_or", "mlx.core.bitwise_xor", "mlx.core.block_masked_mm", "mlx.core.broadcast_to", "mlx.core.ceil", "mlx.core.clip", "mlx.core.compile", "mlx.core.concatenate", "mlx.core.conj", "mlx.core.conjugate", "mlx.core.conv1d", "mlx.core.conv2d", "mlx.core.conv3d", "mlx.core.conv_general", "mlx.core.conv_transpose1d", "mlx.core.conv_transpose2d", "mlx.core.conv_transpose3d", "mlx.core.convolve", "mlx.core.cos", "mlx.core.cosh", "mlx.core.cummax", "mlx.core.cummin", "mlx.core.cumprod", "mlx.core.cumsum", "mlx.core.custom_function", "mlx.core.default_device", "mlx.core.default_stream", "mlx.core.degrees", "mlx.core.dequantize", "mlx.core.diag", "mlx.core.diagonal", "mlx.core.disable_compile", "mlx.core.distributed.Group", "mlx.core.distributed.all_gather", "mlx.core.distributed.all_sum", "mlx.core.distributed.init", "mlx.core.distributed.is_available", "mlx.core.distributed.recv", "mlx.core.distributed.recv_like", "mlx.core.distributed.send", "mlx.core.divide", "mlx.core.divmod", "mlx.core.einsum", "mlx.core.einsum_path", "mlx.core.enable_compile", "mlx.core.equal", "mlx.core.erf", "mlx.core.erfinv", "mlx.core.eval", "mlx.core.exp", "mlx.core.expand_dims", "mlx.core.expm1", "mlx.core.eye", "mlx.core.fast.affine_quantize", "mlx.core.fast.layer_norm", "mlx.core.fast.metal_kernel", "mlx.core.fast.rms_norm", "mlx.core.fast.rope", "mlx.core.fast.scaled_dot_product_attention", "mlx.core.fft.fft", "mlx.core.fft.fft2", "mlx.core.fft.fftn", "mlx.core.fft.ifft", "mlx.core.fft.ifft2", "mlx.core.fft.ifftn", "mlx.core.fft.irfft", "mlx.core.fft.irfft2", "mlx.core.fft.irfftn", "mlx.core.fft.rfft", "mlx.core.fft.rfft2", "mlx.core.fft.rfftn", "mlx.core.flatten", "mlx.core.floor", "mlx.core.floor_divide", "mlx.core.full", "mlx.core.gather_mm", "mlx.core.gather_qmm", "mlx.core.grad", "mlx.core.greater", "mlx.core.greater_equal", "mlx.core.hadamard_transform", "mlx.core.identity", "mlx.core.imag", "mlx.core.inner", "mlx.core.isclose", "mlx.core.isfinite", "mlx.core.isinf", "mlx.core.isnan", "mlx.core.isneginf", "mlx.core.isposinf", "mlx.core.issubdtype", "mlx.core.jvp", "mlx.core.left_shift", "mlx.core.less", "mlx.core.less_equal", "mlx.core.linalg.cholesky", "mlx.core.linalg.cholesky_inv", "mlx.core.linalg.cross", "mlx.core.linalg.eigh", "mlx.core.linalg.eigvalsh", "mlx.core.linalg.inv", "mlx.core.linalg.norm", "mlx.core.linalg.qr", "mlx.core.linalg.svd", "mlx.core.linalg.tri_inv", "mlx.core.linspace", "mlx.core.load", "mlx.core.log", "mlx.core.log10", "mlx.core.log1p", "mlx.core.log2", "mlx.core.logaddexp", "mlx.core.logical_and", "mlx.core.logical_not", "mlx.core.logical_or", "mlx.core.logsumexp", "mlx.core.matmul", "mlx.core.max", "mlx.core.maximum", "mlx.core.mean", "mlx.core.meshgrid", "mlx.core.metal.clear_cache", "mlx.core.metal.device_info", "mlx.core.metal.get_active_memory", "mlx.core.metal.get_cache_memory", "mlx.core.metal.get_peak_memory", "mlx.core.metal.is_available", "mlx.core.metal.reset_peak_memory", "mlx.core.metal.set_cache_limit", "mlx.core.metal.set_memory_limit", "mlx.core.metal.set_wired_limit", "mlx.core.metal.start_capture", "mlx.core.metal.stop_capture", "mlx.core.min", "mlx.core.minimum", "mlx.core.moveaxis", "mlx.core.multiply", "mlx.core.nan_to_num", "mlx.core.negative", "mlx.core.new_stream", "mlx.core.not_equal", "mlx.core.ones", "mlx.core.ones_like", "mlx.core.outer", "mlx.core.pad", "mlx.core.partition", "mlx.core.power", "mlx.core.prod", "mlx.core.put_along_axis", "mlx.core.quantize", "mlx.core.quantized_matmul", "mlx.core.radians", "mlx.core.random.bernoulli", "mlx.core.random.categorical", "mlx.core.random.gumbel", "mlx.core.random.key", "mlx.core.random.laplace", "mlx.core.random.multivariate_normal", "mlx.core.random.normal", "mlx.core.random.permutation", "mlx.core.random.randint", "mlx.core.random.seed", "mlx.core.random.split", "mlx.core.random.truncated_normal", "mlx.core.random.uniform", "mlx.core.real", "mlx.core.reciprocal", "mlx.core.remainder", "mlx.core.repeat", "mlx.core.reshape", "mlx.core.right_shift", "mlx.core.roll", "mlx.core.round", "mlx.core.rsqrt", "mlx.core.save", "mlx.core.save_gguf", "mlx.core.save_safetensors", "mlx.core.savez", "mlx.core.savez_compressed", "mlx.core.set_default_device", "mlx.core.set_default_stream", "mlx.core.sigmoid", "mlx.core.sign", "mlx.core.sin", "mlx.core.sinh", "mlx.core.softmax", "mlx.core.sort", "mlx.core.split", "mlx.core.sqrt", "mlx.core.square", "mlx.core.squeeze", "mlx.core.stack", "mlx.core.std", "mlx.core.stop_gradient", "mlx.core.stream", "mlx.core.subtract", "mlx.core.sum", "mlx.core.swapaxes", "mlx.core.synchronize", "mlx.core.take", "mlx.core.take_along_axis", "mlx.core.tan", "mlx.core.tanh", "mlx.core.tensordot", "mlx.core.tile", "mlx.core.topk", "mlx.core.trace", "mlx.core.transpose", "mlx.core.tri", "mlx.core.tril", "mlx.core.triu", "mlx.core.value_and_grad", "mlx.core.var", "mlx.core.view", "mlx.core.vjp", "mlx.core.vmap", "mlx.core.where", "mlx.core.zeros", "mlx.core.zeros_like", "mlx.nn.quantize", "mlx.nn.value_and_grad", "mlx.optimizers.clip_grad_norm", "mlx.utils.tree_flatten", "mlx.utils.tree_map", "mlx.utils.tree_map_with_path", "mlx.utils.tree_reduce", "mlx.utils.tree_unflatten", "mlx.core.Stream", "Array", "Data Types", "Devices and Streams", "Distributed Communication", "Fast", "FFT", "Linear Algebra", "Metal", "Neural Networks", "mlx.nn.ALiBi", "mlx.nn.AvgPool1d", "mlx.nn.AvgPool2d", "mlx.nn.BatchNorm", "mlx.nn.CELU", "mlx.nn.Conv1d", "mlx.nn.Conv2d", "mlx.nn.Conv3d", "mlx.nn.ConvTranspose1d", "mlx.nn.ConvTranspose2d", "mlx.nn.ConvTranspose3d", "mlx.nn.Dropout", "mlx.nn.Dropout2d", "mlx.nn.Dropout3d", "mlx.nn.ELU", "mlx.nn.Embedding", "mlx.nn.GELU", "mlx.nn.GLU", "mlx.nn.GRU", "mlx.nn.GroupNorm", "mlx.nn.HardShrink", "mlx.nn.HardTanh", "mlx.nn.Hardswish", "mlx.nn.InstanceNorm", "mlx.nn.LSTM", "mlx.nn.LayerNorm", "mlx.nn.LeakyReLU", "mlx.nn.Linear", "mlx.nn.LogSigmoid", "mlx.nn.LogSoftmax", "mlx.nn.MaxPool1d", "mlx.nn.MaxPool2d", "mlx.nn.Mish", "mlx.nn.Module.apply", "mlx.nn.Module.apply_to_modules", "mlx.nn.Module.children", "mlx.nn.Module.eval", "mlx.nn.Module.filter_and_map", "mlx.nn.Module.freeze", "mlx.nn.Module.leaf_modules", "mlx.nn.Module.load_weights", "mlx.nn.Module.modules", "mlx.nn.Module.named_modules", "mlx.nn.Module.parameters", "mlx.nn.Module.save_weights", "mlx.nn.Module.set_dtype", "mlx.nn.Module.state", "mlx.nn.Module.train", "mlx.nn.Module.trainable_parameters", "mlx.nn.Module.training", "mlx.nn.Module.unfreeze", "mlx.nn.Module.update", "mlx.nn.Module.update_modules", "mlx.nn.MultiHeadAttention", "mlx.nn.PReLU", "mlx.nn.QuantizedEmbedding", "mlx.nn.QuantizedLinear", "mlx.nn.RMSNorm", "mlx.nn.RNN", "mlx.nn.ReLU", "mlx.nn.ReLU6", "mlx.nn.RoPE", "mlx.nn.SELU", "mlx.nn.Sequential", "mlx.nn.SiLU", "mlx.nn.Sigmoid", "mlx.nn.SinusoidalPositionalEncoding", "mlx.nn.Softmax", "mlx.nn.Softmin", "mlx.nn.Softplus", "mlx.nn.Softshrink", "mlx.nn.Softsign", "mlx.nn.Step", "mlx.nn.Tanh", "mlx.nn.Transformer", "mlx.nn.Upsample", "mlx.nn.init.constant", "mlx.nn.init.glorot_normal", "mlx.nn.init.glorot_uniform", "mlx.nn.init.he_normal", "mlx.nn.init.he_uniform", "mlx.nn.init.identity", "mlx.nn.init.normal", "mlx.nn.init.uniform", "mlx.nn.celu", "mlx.nn.elu", "mlx.nn.gelu", "mlx.nn.gelu_approx", "mlx.nn.gelu_fast_approx", "mlx.nn.glu", "mlx.nn.hard_shrink", "mlx.nn.hard_tanh", "mlx.nn.hardswish", "mlx.nn.leaky_relu", "mlx.nn.log_sigmoid", "mlx.nn.log_softmax", "mlx.nn.losses.binary_cross_entropy", "mlx.nn.losses.cosine_similarity_loss", "mlx.nn.losses.cross_entropy", "mlx.nn.losses.gaussian_nll_loss", "mlx.nn.losses.hinge_loss", "mlx.nn.losses.huber_loss", "mlx.nn.losses.kl_div_loss", "mlx.nn.losses.l1_loss", "mlx.nn.losses.log_cosh_loss", "mlx.nn.losses.margin_ranking_loss", "mlx.nn.losses.mse_loss", "mlx.nn.losses.nll_loss", "mlx.nn.losses.smooth_l1_loss", "mlx.nn.losses.triplet_loss", "mlx.nn.mish", "mlx.nn.prelu", "mlx.nn.relu", "mlx.nn.relu6", "mlx.nn.selu", "mlx.nn.sigmoid", "mlx.nn.silu", "mlx.nn.softmax", "mlx.nn.softmin", "mlx.nn.softplus", "mlx.nn.softshrink", "mlx.nn.step", "mlx.nn.tanh", "Functions", "Initializers", "Layers", "Loss Functions", "Module", "Operations", "Optimizers", "mlx.optimizers.AdaDelta", "mlx.optimizers.Adafactor", "mlx.optimizers.Adagrad", "mlx.optimizers.Adam", "mlx.optimizers.AdamW", "mlx.optimizers.Adamax", "mlx.optimizers.Lion", "mlx.optimizers.Optimizer.apply_gradients", "mlx.optimizers.Optimizer.init", "mlx.optimizers.Optimizer.state", "mlx.optimizers.Optimizer.update", "mlx.optimizers.RMSprop", "mlx.optimizers.SGD", "mlx.optimizers.cosine_decay", "mlx.optimizers.exponential_decay", "mlx.optimizers.join_schedules", "mlx.optimizers.linear_schedule", "mlx.optimizers.step_decay", "Common Optimizers", "Optimizer", "Schedulers", "Random", "Transforms", "Tree Utils", "Compilation", "Distributed Communication", "Function Transforms", "Indexing Arrays", "Lazy Evaluation", "Conversion to NumPy and Other Frameworks", "Quick Start Guide", "Saving and Loading Arrays", "Unified Memory", "Using Streams"], "titleterms": {"A": 487, "In": 482, "The": 324, "ab": [12, 32], "adadelta": 455, "adafactor": 456, "adagrad": 457, "adam": 458, "adamax": 460, "adamw": 459, "add": 13, "addmm": 14, "affine_quant": 141, "algebra": 322, "alibi": 325, "all": [5, 15, 33, 480], "all_gath": 121, "all_sum": 122, "allclos": 16, "ani": [17, 34], "api": [7, 8], "appli": 358, "apply_gradi": 462, "apply_to_modul": 359, "arang": 18, "arcco": 19, "arccosh": 20, "arcsin": 21, "arcsinh": 22, "arctan": 23, "arctan2": 24, "arctanh": 25, "argmax": [26, 35], "argmin": [27, 36], "argpartit": 28, "argsort": 29, "arrai": [30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 316, 482, 486], "array_equ": 82, "as_strid": 83, "astyp": 37, "atleast_1d": 84, "atleast_2d": 85, "atleast_3d": 86, "attent": 5, "automat": 481, "avgpool1d": 326, "avgpool2d": 327, "back": 2, "basic": [479, 485], "batchnorm": 328, "benchmark": 5, "bernoulli": 240, "binari": 8, "binary_cross_entropi": 421, "bind": 2, "bitwise_and": 87, "bitwise_or": 88, "bitwise_xor": 89, "block_masked_mm": 90, "broadcast_to": 91, "build": [2, 8], "c": [7, 8], "categor": 241, "ceil": 92, "celu": [329, 409], "children": 360, "choleski": 183, "cholesky_inv": 184, "class": 324, "clear_cach": 209, "clip": 93, "clip_grad_norm": 309, "cmake": 2, "co": [40, 106], "code": [2, 5], "common": 473, "commun": [319, 480], "compil": [94, 479], "complex": 1, "comput": 483, "concaten": 95, "conj": [39, 96], "conjug": 97, "constant": 401, "conv1d": [98, 330], "conv2d": [99, 331], "conv3d": [100, 332], "conv_gener": 101, "conv_transpose1d": 102, "conv_transpose2d": 103, "conv_transpose3d": 104, "convers": 484, "convert": 5, "convolv": 105, "convtranspose1d": 333, "convtranspose2d": 334, "convtranspose3d": 335, "core": [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 267, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 315], "cosh": 107, "cosine_decai": 468, "cosine_similarity_loss": 422, "cpu": 2, "cross": 185, "cross_entropi": 423, "cummax": [41, 108], "cummin": [42, 109], "cumprod": [43, 110], "cumsum": [44, 111], "custom": [1, 2], "custom_funct": 112, "data": 317, "debug": 479, "debugg": 3, "default_devic": 113, "default_stream": 114, "degre": 115, "dequant": 116, "devic": [9, 318], "device_info": 210, "diag": [45, 117], "diagon": [46, 118], "differ": 482, "differenti": 481, "disable_compil": 119, "distribut": [120, 121, 122, 123, 124, 125, 126, 127, 319, 480], "divid": 128, "divmod": 129, "download": [2, 5], "dropout": 336, "dropout2d": 337, "dropout3d": 338, "dtype": [10, 47], "dtypecategori": 11, "eigh": 186, "eigvalsh": 187, "einsum": 130, "einsum_path": 131, "elu": [339, 410], "embed": 340, "enable_compil": 132, "encod": 5, "end": 2, "equal": 133, "erf": 134, "erfinv": 135, "eval": [136, 361], "evalu": 483, "exampl": [1, 2, 7, 479, 480, 487], "exp": [48, 137], "expand_dim": 138, "expm1": 139, "exponential_decai": 469, "extens": 2, "ey": 140, "fast": [141, 142, 143, 144, 145, 146, 320], "fft": [147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 321], "fft2": 148, "fftn": 149, "filter_and_map": 362, "flatten": [49, 159], "floor": 160, "floor_divid": 161, "format": 486, "found": 8, "framework": 484, "freez": 363, "from": [8, 482], "full": [5, 162], "function": [448, 451, 479, 481, 485], "further": 7, "gather_mm": 163, "gather_qmm": 164, "gaussian_nll_loss": 424, "gelu": [341, 411], "gelu_approx": 412, "gelu_fast_approx": 413, "gener": 5, "get": 480, "get_active_memori": 211, "get_cache_memori": 212, "get_peak_memori": 213, "glorot_norm": 402, "glorot_uniform": 403, "glu": [342, 414], "gpu": 2, "grad": [165, 324], "graph": [479, 483, 485], "greater": 166, "greater_equ": 167, "grid": 1, "group": 120, "groupnorm": 344, "gru": 343, "guid": 485, "gumbel": 242, "hadamard_transform": 168, "hard_shrink": 415, "hard_tanh": 416, "hardshrink": 345, "hardswish": [347, 417], "hardtanh": 346, "he_norm": 404, "he_uniform": 405, "hinge_loss": 425, "host": 480, "huber_loss": 426, "ident": [169, 406], "ifft": 150, "ifft2": 151, "ifftn": 152, "imag": 170, "implement": [2, 5], "index": 482, "infer": 5, "init": [123, 401, 402, 403, 404, 405, 406, 407, 408, 463], "initi": 449, "inner": 171, "inspect": 324, "instal": [7, 8, 480], "instancenorm": 348, "introduc": 2, "inv": 188, "irfft": 153, "irfft2": 154, "irfftn": 155, "is_avail": [124, 214], "isclos": 172, "isfinit": 173, "isinf": 174, "isnan": 175, "isneginf": 176, "isposinf": 177, "issubdtyp": 178, "item": 50, "items": 51, "jax": 484, "join_schedul": 470, "jvp": 179, "kei": 243, "kernel": 1, "kl_div_loss": 427, "l1_loss": 428, "laplac": 244, "layer": [5, 6, 450], "layer_norm": 142, "layernorm": 350, "lazi": 483, "leaf_modul": 364, "leaky_relu": 418, "leakyrelu": 351, "left_shift": 180, "less": 181, "less_equ": 182, "linalg": [183, 184, 185, 186, 187, 188, 189, 190, 191, 192], "linear": [4, 322, 352], "linear_schedul": 471, "linspac": 193, "lion": 461, "llm": 5, "load": [5, 194, 454, 486], "load_weight": 365, "log": [52, 195], "log10": [53, 196], "log1p": [54, 197], "log2": [55, 198], "log_cosh_loss": 429, "log_sigmoid": 419, "log_softmax": 420, "logaddexp": 199, "logical_and": 200, "logical_not": 201, "logical_or": 202, "logsigmoid": 353, "logsoftmax": 354, "logsumexp": [56, 203], "loss": [421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 451], "lstm": 349, "margin_ranking_loss": 430, "matmul": 204, "max": [57, 205], "maximum": 206, "maxpool1d": 355, "maxpool2d": 356, "mean": [58, 207], "memori": 487, "meshgrid": 208, "metal": [1, 3, 8, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 323], "metal_kernel": 143, "min": [59, 221], "minim": 8, "minimum": 222, "mish": [357, 435], "mlx": [2, 7, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 267, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 307, 308, 309, 310, 311, 312, 313, 314, 315, 325, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350, 351, 352, 353, 354, 355, 356, 357, 358, 359, 360, 361, 362, 363, 364, 365, 366, 367, 368, 369, 370, 371, 372, 373, 374, 375, 376, 377, 378, 379, 380, 381, 382, 383, 384, 385, 386, 387, 388, 389, 390, 391, 392, 393, 394, 395, 396, 397, 398, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 409, 410, 411, 412, 413, 414, 415, 416, 417, 418, 419, 420, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445, 446, 447, 455, 456, 457, 458, 459, 460, 461, 462, 463, 464, 465, 466, 467, 468, 469, 470, 471, 472], "model": 5, "modul": [324, 358, 359, 360, 361, 362, 363, 364, 365, 366, 367, 368, 369, 370, 371, 372, 373, 374, 375, 376, 377, 452], "moveaxi": [60, 223], "mpi": 480, "mse_loss": 431, "multi": 6, "multiheadattent": 378, "multipli": 224, "multivariate_norm": 245, "named_modul": 367, "nan_to_num": 225, "nbyte": 61, "ndim": 62, "neg": 226, "network": 324, "neural": 324, "new_stream": 227, "nll_loss": 432, "nn": [307, 308, 325, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350, 351, 352, 353, 354, 355, 356, 357, 358, 359, 360, 361, 362, 363, 364, 365, 366, 367, 368, 369, 370, 371, 372, 373, 374, 375, 376, 377, 378, 379, 380, 381, 382, 383, 384, 385, 386, 387, 388, 389, 390, 391, 392, 393, 394, 395, 396, 397, 398, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 409, 410, 411, 412, 413, 414, 415, 416, 417, 418, 419, 420, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445, 446, 447], "norm": 189, "normal": [246, 407], "not_equ": 228, "numpi": [482, 484], "ones": 229, "ones_lik": 230, "onli": 483, "oper": [0, 2, 453], "optim": [309, 454, 455, 456, 457, 458, 459, 460, 461, 462, 463, 464, 465, 466, 467, 468, 469, 470, 471, 472, 473, 474], "option": 8, "other": 484, "outer": 231, "pad": 232, "paramet": [324, 368], "partit": 233, "perceptron": 6, "permut": 247, "place": 482, "power": 234, "prelu": [379, 436], "primit": 2, "prod": [63, 235], "pure": 479, "put": 5, "put_along_axi": 236, "python": [2, 7, 8], "pytorch": 484, "qr": 190, "quantiz": [237, 307], "quantized_matmul": 238, "quantizedembed": 380, "quantizedlinear": 381, "quick": [324, 485], "radian": 239, "randint": 248, "random": [240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 476], "read": 7, "real": 253, "reciproc": [64, 254], "recv": 125, "recv_lik": 126, "reduc": 480, "refer": 7, "regress": 4, "relu": [384, 437], "relu6": [385, 438], "remaind": 255, "remot": 480, "repeat": 256, "requir": 8, "reset_peak_memori": 215, "reshap": [65, 257], "result": 2, "rfft": 156, "rfft2": 157, "rfftn": 158, "right_shift": 258, "rms_norm": 144, "rmsnorm": 382, "rmsprop": 466, "rnn": 383, "roll": 259, "rope": [145, 386], "round": [66, 260], "rsqrt": [67, 261], "sampl": 1, "save": [262, 454, 486], "save_gguf": 263, "save_safetensor": 264, "save_weight": 369, "savez": 265, "savez_compress": 266, "scaled_dot_product_attent": 146, "schedul": 475, "script": [2, 5], "seed": 249, "selu": [387, 439], "send": 127, "sequenti": 388, "serial": 486, "set": 480, "set_cache_limit": 216, "set_default_devic": 267, "set_default_stream": 268, "set_dtyp": 370, "set_memory_limit": 217, "set_wired_limit": 218, "setuptool": 2, "sgd": 467, "shape": [1, 68], "shell": 8, "sigmoid": [269, 390, 440], "sign": 270, "silu": [389, 441], "simpl": [1, 487], "sin": [69, 271], "sinh": 272, "sinusoidalpositionalencod": 391, "size": [8, 70], "smooth_l1_loss": 433, "softmax": [273, 392, 442], "softmin": [393, 443], "softplu": [394, 444], "softshrink": [395, 445], "softsign": 396, "sort": 274, "sourc": 8, "specifi": 488, "speedup": 479, "split": [71, 250, 275], "sqrt": [72, 276], "squar": [73, 277], "squeez": [74, 278], "stack": 279, "start": [324, 480, 485], "start_captur": 219, "state": [371, 464], "std": [75, 280], "step": [397, 446], "step_decai": 472, "stop_captur": 220, "stop_gradi": 281, "stream": [282, 315, 318, 488], "stride": 1, "subtract": 283, "sum": [76, 284], "support": 317, "svd": 191, "swapax": [77, 285], "synchron": 286, "t": 31, "take": 287, "take_along_axi": 288, "tan": 289, "tanh": [290, 398, 447], "tensordot": 291, "tensorflow": 484, "tile": 292, "togeth": 5, "tolist": 78, "topk": 293, "trace": 294, "train": [372, 374, 479, 480], "trainable_paramet": 373, "transform": [2, 399, 477, 479, 481, 483, 485], "transpos": [79, 295], "tree": 478, "tree_flatten": 310, "tree_map": 311, "tree_map_with_path": 312, "tree_reduc": 313, "tree_unflatten": 314, "tri": 296, "tri_inv": 192, "tril": 297, "triplet_loss": 434, "triu": 298, "troubleshoot": 8, "truncated_norm": 251, "tune": 480, "type": 317, "unfreez": 375, "unifi": 487, "uniform": [252, 408], "up": 480, "updat": [324, 376, 465, 482], "update_modul": 377, "upsampl": 400, "us": [1, 2, 483, 488], "usag": [2, 7], "util": [310, 311, 312, 313, 314, 478], "valu": 324, "value_and_grad": [299, 308], "var": [80, 300], "vector": 481, "view": [81, 301], "vjp": [1, 302], "vmap": 303, "weight": 5, "what": 483, "when": 483, "where": 304, "why": 483, "workflow": 3, "x86": 8, "xcode": 3, "you": 483, "zero": 305, "zeros_lik": 306}}) \ No newline at end of file +Search.setIndex({"alltitles": {"A Simple Example": [[487, "a-simple-example"]], "Array": [[316, null]], "Attention layer": [[5, "attention-layer"]], "Automatic Differentiation": [[481, "automatic-differentiation"]], "Automatic Vectorization": [[481, "automatic-vectorization"]], "Basics": [[485, "basics"]], "Basics of Compile": [[479, "basics-of-compile"]], "Binary Size Minimization": [[8, "binary-size-minimization"]], "Binding to Python": [[2, "binding-to-python"]], "Build Options": [[8, "id3"]], "Build Requirements": [[8, "build-requirements"]], "Build and Install": [[8, null]], "Build from source": [[8, "build-from-source"]], "Building and Binding": [[2, "building-and-binding"]], "Building with CMake": [[2, "building-with-cmake"]], "Building with setuptools": [[2, "building-with-setuptools"]], "C++ API": [[8, "c-api"]], "C++ API Reference": [[7, null]], "Common Optimizers": [[473, null]], "Compilation": [[479, null]], "Compiling Training Graphs": [[479, "compiling-training-graphs"]], "Complex Example": [[1, "complex-example"]], "Conversion to NumPy and Other Frameworks": [[484, null]], "Converting the weights": [[5, "converting-the-weights"]], "Custom Extensions in MLX": [[2, null]], "Custom Metal Kernels": [[1, null]], "Data Types": [[317, null]], "Debugging": [[479, "debugging"]], "Devices and Streams": [[318, null]], "Differences from NumPy": [[482, "differences-from-numpy"]], "Distributed Communication": [[319, null], [480, null]], "Download the code": [[2, null], [5, null]], "Encoder layer": [[5, "encoder-layer"]], "Example Speedup": [[479, "example-speedup"]], "Examples": [[7, null]], "FFT": [[321, null]], "Fast": [[320, null]], "Full model": [[5, "full-model"]], "Function Transforms": [[481, null]], "Function and Graph Transformations": [[485, "function-and-graph-transformations"]], "Functions": [[448, null]], "Further Reading": [[7, null]], "Generation": [[5, "generation"]], "Getting Started": [[480, "getting-started"]], "Grid Sample VJP": [[1, "grid-sample-vjp"]], "Implementing the CPU Back-end": [[2, "implementing-the-cpu-back-end"]], "Implementing the GPU Back-end": [[2, "implementing-the-gpu-back-end"]], "Implementing the Primitive": [[2, "implementing-the-primitive"]], "Implementing the model": [[5, "implementing-the-model"]], "In Place Updates": [[482, "in-place-updates"]], "Indexing Arrays": [[482, null]], "Initializers": [[449, null]], "Inspecting Modules": [[324, "inspecting-modules"]], "Install": [[7, null]], "Installing MPI": [[480, "installing-mpi"]], "Introducing the Example": [[2, "introducing-the-example"]], "JAX": [[484, "jax"]], "LLM inference": [[5, null]], "Layers": [[450, null]], "Lazy Evaluation": [[483, null]], "Linear Algebra": [[322, null]], "Linear Regression": [[4, null]], "Loss Functions": [[451, null]], "MLX": [[7, null]], "Metal": [[323, null]], "Metal Debugger": [[3, null]], "Metal not found": [[8, "metal-not-found"]], "Module": [[452, null]], "Multi-Layer Perceptron": [[6, null]], "Neural Networks": [[324, null]], "Only Compute What You Use": [[483, "only-compute-what-you-use"]], "Operations": [[0, null], [2, "operations"], [453, null]], "Operations and Primitives": [[2, "operations-and-primitives"]], "Optimizer": [[474, null]], "Optimizers": [[454, null]], "Parameters": [[324, "parameters"]], "Primitive Transforms": [[2, "primitive-transforms"]], "Primitives": [[2, "primitives"]], "Pure Functions": [[479, "pure-functions"]], "Putting it all together": [[5, "putting-it-all-together"]], "PyTorch": [[484, "pytorch"]], "Python API": [[8, "python-api"]], "Python API Reference": [[7, null]], "Python Installation": [[8, "python-installation"]], "Quick Start Guide": [[485, null]], "Quick Start with Neural Networks": [[324, "quick-start-with-neural-networks"]], "Random": [[476, null]], "Results": [[2, "results"]], "Saving and Loading": [[454, "saving-and-loading"]], "Saving and Loading Arrays": [[486, null]], "Schedulers": [[475, null]], "Scripts": [[2, "scripts"], [5, "scripts"]], "Serialization Formats": [[486, "id1"]], "Setting up Remote Hosts": [[480, "setting-up-remote-hosts"]], "Simple Example": [[1, "simple-example"]], "Specifying the Stream": [[488, "specifying-the-stream"]], "Supported Data Types": [[317, "id2"]], "TensorFlow": [[484, "tensorflow"]], "The Module Class": [[324, "the-module-class"]], "Training Example": [[480, "training-example"]], "Transformations with Compile": [[479, "transformations-with-compile"]], "Transforming Compute Graphs": [[483, "transforming-compute-graphs"]], "Transforms": [[477, null]], "Tree Utils": [[478, null]], "Troubleshooting": [[8, "troubleshooting"], [8, "id2"]], "Tuning All Reduce": [[480, "tuning-all-reduce"]], "Unified Memory": [[487, null]], "Updating the Parameters": [[324, "updating-the-parameters"]], "Usage": [[2, "usage"], [7, null]], "Using Shape/Strides": [[1, "using-shape-strides"]], "Using Streams": [[488, null]], "Using the Primitive": [[2, "using-the-primitive"]], "Value and Grad": [[324, "value-and-grad"]], "Weight loading and benchmarking": [[5, "weight-loading-and-benchmarking"]], "When to Evaluate": [[483, "when-to-evaluate"]], "Why Lazy Evaluation": [[483, "why-lazy-evaluation"]], "Xcode Workflow": [[3, "xcode-workflow"]], "mlx.core.Device": [[9, null]], "mlx.core.Dtype": [[10, null]], "mlx.core.DtypeCategory": [[11, null]], "mlx.core.Stream": [[315, null]], "mlx.core.abs": [[12, null]], "mlx.core.add": [[13, null]], "mlx.core.addmm": [[14, null]], "mlx.core.all": [[15, null]], "mlx.core.allclose": [[16, null]], "mlx.core.any": [[17, null]], "mlx.core.arange": [[18, null]], "mlx.core.arccos": [[19, null]], "mlx.core.arccosh": [[20, null]], "mlx.core.arcsin": [[21, null]], "mlx.core.arcsinh": [[22, null]], "mlx.core.arctan": [[23, null]], "mlx.core.arctan2": [[24, null]], "mlx.core.arctanh": [[25, null]], "mlx.core.argmax": [[26, null]], "mlx.core.argmin": [[27, null]], "mlx.core.argpartition": [[28, null]], "mlx.core.argsort": [[29, null]], "mlx.core.array": [[30, null]], "mlx.core.array.T": [[31, null]], "mlx.core.array.abs": [[32, null]], "mlx.core.array.all": [[33, null]], "mlx.core.array.any": [[34, null]], "mlx.core.array.argmax": [[35, null]], "mlx.core.array.argmin": [[36, null]], "mlx.core.array.astype": [[37, null]], "mlx.core.array.at": [[38, null]], "mlx.core.array.conj": [[39, null]], "mlx.core.array.cos": [[40, null]], "mlx.core.array.cummax": [[41, null]], "mlx.core.array.cummin": [[42, null]], "mlx.core.array.cumprod": [[43, null]], "mlx.core.array.cumsum": [[44, null]], "mlx.core.array.diag": [[45, null]], "mlx.core.array.diagonal": [[46, null]], "mlx.core.array.dtype": [[47, null]], "mlx.core.array.exp": [[48, null]], "mlx.core.array.flatten": [[49, null]], "mlx.core.array.item": [[50, null]], "mlx.core.array.itemsize": [[51, null]], "mlx.core.array.log": [[52, null]], "mlx.core.array.log10": [[53, null]], "mlx.core.array.log1p": [[54, null]], "mlx.core.array.log2": [[55, null]], "mlx.core.array.logsumexp": [[56, null]], "mlx.core.array.max": [[57, null]], "mlx.core.array.mean": [[58, null]], "mlx.core.array.min": [[59, null]], "mlx.core.array.moveaxis": [[60, null]], "mlx.core.array.nbytes": [[61, null]], "mlx.core.array.ndim": [[62, null]], "mlx.core.array.prod": [[63, null]], "mlx.core.array.reciprocal": [[64, null]], "mlx.core.array.reshape": [[65, null]], "mlx.core.array.round": [[66, null]], "mlx.core.array.rsqrt": [[67, null]], "mlx.core.array.shape": [[68, null]], "mlx.core.array.sin": [[69, null]], "mlx.core.array.size": [[70, null]], "mlx.core.array.split": [[71, null]], "mlx.core.array.sqrt": [[72, null]], "mlx.core.array.square": [[73, null]], "mlx.core.array.squeeze": [[74, null]], "mlx.core.array.std": [[75, null]], "mlx.core.array.sum": [[76, null]], "mlx.core.array.swapaxes": [[77, null]], "mlx.core.array.tolist": [[78, null]], "mlx.core.array.transpose": [[79, null]], "mlx.core.array.var": [[80, null]], "mlx.core.array.view": [[81, null]], "mlx.core.array_equal": [[82, null]], "mlx.core.as_strided": [[83, null]], "mlx.core.atleast_1d": [[84, null]], "mlx.core.atleast_2d": [[85, null]], "mlx.core.atleast_3d": [[86, null]], "mlx.core.bitwise_and": [[87, null]], "mlx.core.bitwise_or": [[88, null]], "mlx.core.bitwise_xor": [[89, null]], "mlx.core.block_masked_mm": [[90, null]], "mlx.core.broadcast_to": [[91, null]], "mlx.core.ceil": [[92, null]], "mlx.core.clip": [[93, null]], "mlx.core.compile": [[94, null]], "mlx.core.concatenate": [[95, null]], "mlx.core.conj": [[96, null]], "mlx.core.conjugate": [[97, null]], "mlx.core.conv1d": [[98, null]], "mlx.core.conv2d": [[99, null]], "mlx.core.conv3d": [[100, null]], "mlx.core.conv_general": [[101, null]], "mlx.core.conv_transpose1d": [[102, null]], "mlx.core.conv_transpose2d": [[103, null]], "mlx.core.conv_transpose3d": [[104, null]], "mlx.core.convolve": [[105, null]], "mlx.core.cos": [[106, null]], "mlx.core.cosh": [[107, null]], "mlx.core.cummax": [[108, null]], "mlx.core.cummin": [[109, null]], "mlx.core.cumprod": [[110, null]], "mlx.core.cumsum": [[111, null]], "mlx.core.custom_function": [[112, null]], "mlx.core.default_device": [[113, null]], "mlx.core.default_stream": [[114, null]], "mlx.core.degrees": [[115, null]], "mlx.core.dequantize": [[116, null]], "mlx.core.diag": [[117, null]], "mlx.core.diagonal": [[118, null]], "mlx.core.disable_compile": [[119, null]], "mlx.core.distributed.Group": [[120, null]], "mlx.core.distributed.all_gather": [[121, null]], "mlx.core.distributed.all_sum": [[122, null]], "mlx.core.distributed.init": [[123, null]], "mlx.core.distributed.is_available": [[124, null]], "mlx.core.distributed.recv": [[125, null]], "mlx.core.distributed.recv_like": [[126, null]], "mlx.core.distributed.send": [[127, null]], "mlx.core.divide": [[128, null]], "mlx.core.divmod": [[129, null]], "mlx.core.einsum": [[130, null]], "mlx.core.einsum_path": [[131, null]], "mlx.core.enable_compile": [[132, null]], "mlx.core.equal": [[133, null]], "mlx.core.erf": [[134, null]], "mlx.core.erfinv": [[135, null]], "mlx.core.eval": [[136, null]], "mlx.core.exp": [[137, null]], "mlx.core.expand_dims": [[138, null]], "mlx.core.expm1": [[139, null]], "mlx.core.eye": [[140, null]], "mlx.core.fast.affine_quantize": [[141, null]], "mlx.core.fast.layer_norm": [[142, null]], "mlx.core.fast.metal_kernel": [[143, null]], "mlx.core.fast.rms_norm": [[144, null]], "mlx.core.fast.rope": [[145, null]], "mlx.core.fast.scaled_dot_product_attention": [[146, null]], "mlx.core.fft.fft": [[147, null]], "mlx.core.fft.fft2": [[148, null]], "mlx.core.fft.fftn": [[149, null]], "mlx.core.fft.ifft": [[150, null]], "mlx.core.fft.ifft2": [[151, null]], "mlx.core.fft.ifftn": [[152, null]], "mlx.core.fft.irfft": [[153, null]], "mlx.core.fft.irfft2": [[154, null]], "mlx.core.fft.irfftn": [[155, null]], "mlx.core.fft.rfft": [[156, null]], "mlx.core.fft.rfft2": [[157, null]], "mlx.core.fft.rfftn": [[158, null]], "mlx.core.flatten": [[159, null]], "mlx.core.floor": [[160, null]], "mlx.core.floor_divide": [[161, null]], "mlx.core.full": [[162, null]], "mlx.core.gather_mm": [[163, null]], "mlx.core.gather_qmm": [[164, null]], "mlx.core.grad": [[165, null]], "mlx.core.greater": [[166, null]], "mlx.core.greater_equal": [[167, null]], "mlx.core.hadamard_transform": [[168, null]], "mlx.core.identity": [[169, null]], "mlx.core.imag": [[170, null]], "mlx.core.inner": [[171, null]], "mlx.core.isclose": [[172, null]], "mlx.core.isfinite": [[173, null]], "mlx.core.isinf": [[174, null]], "mlx.core.isnan": [[175, null]], "mlx.core.isneginf": [[176, null]], "mlx.core.isposinf": [[177, null]], "mlx.core.issubdtype": [[178, null]], "mlx.core.jvp": [[179, null]], "mlx.core.left_shift": [[180, null]], "mlx.core.less": [[181, null]], "mlx.core.less_equal": [[182, null]], "mlx.core.linalg.cholesky": [[183, null]], "mlx.core.linalg.cholesky_inv": [[184, null]], "mlx.core.linalg.cross": [[185, null]], "mlx.core.linalg.eigh": [[186, null]], "mlx.core.linalg.eigvalsh": [[187, null]], "mlx.core.linalg.inv": [[188, null]], "mlx.core.linalg.norm": [[189, null]], "mlx.core.linalg.qr": [[190, null]], "mlx.core.linalg.svd": [[191, null]], "mlx.core.linalg.tri_inv": [[192, null]], "mlx.core.linspace": [[193, null]], "mlx.core.load": [[194, null]], "mlx.core.log": [[195, null]], "mlx.core.log10": [[196, null]], "mlx.core.log1p": [[197, null]], "mlx.core.log2": [[198, null]], "mlx.core.logaddexp": [[199, null]], "mlx.core.logical_and": [[200, null]], "mlx.core.logical_not": [[201, null]], "mlx.core.logical_or": [[202, null]], "mlx.core.logsumexp": [[203, null]], "mlx.core.matmul": [[204, null]], "mlx.core.max": [[205, null]], "mlx.core.maximum": [[206, null]], "mlx.core.mean": [[207, null]], "mlx.core.meshgrid": [[208, null]], "mlx.core.metal.clear_cache": [[209, null]], "mlx.core.metal.device_info": [[210, null]], "mlx.core.metal.get_active_memory": [[211, null]], "mlx.core.metal.get_cache_memory": [[212, null]], "mlx.core.metal.get_peak_memory": [[213, null]], "mlx.core.metal.is_available": [[214, null]], "mlx.core.metal.reset_peak_memory": [[215, null]], "mlx.core.metal.set_cache_limit": [[216, null]], "mlx.core.metal.set_memory_limit": [[217, null]], "mlx.core.metal.set_wired_limit": [[218, null]], "mlx.core.metal.start_capture": [[219, null]], "mlx.core.metal.stop_capture": [[220, null]], "mlx.core.min": [[221, null]], "mlx.core.minimum": [[222, null]], "mlx.core.moveaxis": [[223, null]], "mlx.core.multiply": [[224, null]], "mlx.core.nan_to_num": [[225, null]], "mlx.core.negative": [[226, null]], "mlx.core.new_stream": [[227, null]], "mlx.core.not_equal": [[228, null]], "mlx.core.ones": [[229, null]], "mlx.core.ones_like": [[230, null]], "mlx.core.outer": [[231, null]], "mlx.core.pad": [[232, null]], "mlx.core.partition": [[233, null]], "mlx.core.power": [[234, null]], "mlx.core.prod": [[235, null]], "mlx.core.put_along_axis": [[236, null]], "mlx.core.quantize": [[237, null]], "mlx.core.quantized_matmul": [[238, null]], "mlx.core.radians": [[239, null]], "mlx.core.random.bernoulli": [[240, null]], "mlx.core.random.categorical": [[241, null]], "mlx.core.random.gumbel": [[242, null]], "mlx.core.random.key": [[243, null]], "mlx.core.random.laplace": [[244, null]], "mlx.core.random.multivariate_normal": [[245, null]], "mlx.core.random.normal": [[246, null]], "mlx.core.random.permutation": [[247, null]], "mlx.core.random.randint": [[248, null]], "mlx.core.random.seed": [[249, null]], "mlx.core.random.split": [[250, null]], "mlx.core.random.truncated_normal": [[251, null]], "mlx.core.random.uniform": [[252, null]], "mlx.core.real": [[253, null]], "mlx.core.reciprocal": [[254, null]], "mlx.core.remainder": [[255, null]], "mlx.core.repeat": [[256, null]], "mlx.core.reshape": [[257, null]], "mlx.core.right_shift": [[258, null]], "mlx.core.roll": [[259, null]], "mlx.core.round": [[260, null]], "mlx.core.rsqrt": [[261, null]], "mlx.core.save": [[262, null]], "mlx.core.save_gguf": [[263, null]], "mlx.core.save_safetensors": [[264, null]], "mlx.core.savez": [[265, null]], "mlx.core.savez_compressed": [[266, null]], "mlx.core.set_default_device": [[267, null]], "mlx.core.set_default_stream": [[268, null]], "mlx.core.sigmoid": [[269, null]], "mlx.core.sign": [[270, null]], "mlx.core.sin": [[271, null]], "mlx.core.sinh": [[272, null]], "mlx.core.softmax": [[273, null]], "mlx.core.sort": [[274, null]], "mlx.core.split": [[275, null]], "mlx.core.sqrt": [[276, null]], "mlx.core.square": [[277, null]], "mlx.core.squeeze": [[278, null]], "mlx.core.stack": [[279, null]], "mlx.core.std": [[280, null]], "mlx.core.stop_gradient": [[281, null]], "mlx.core.stream": [[282, null]], "mlx.core.subtract": [[283, null]], "mlx.core.sum": [[284, null]], "mlx.core.swapaxes": [[285, null]], "mlx.core.synchronize": [[286, null]], "mlx.core.take": [[287, null]], "mlx.core.take_along_axis": [[288, null]], "mlx.core.tan": [[289, null]], "mlx.core.tanh": [[290, null]], "mlx.core.tensordot": [[291, null]], "mlx.core.tile": [[292, null]], "mlx.core.topk": [[293, null]], "mlx.core.trace": [[294, null]], "mlx.core.transpose": [[295, null]], "mlx.core.tri": [[296, null]], "mlx.core.tril": [[297, null]], "mlx.core.triu": [[298, null]], "mlx.core.value_and_grad": [[299, null]], "mlx.core.var": [[300, null]], "mlx.core.view": [[301, null]], "mlx.core.vjp": [[302, null]], "mlx.core.vmap": [[303, null]], "mlx.core.where": [[304, null]], "mlx.core.zeros": [[305, null]], "mlx.core.zeros_like": [[306, null]], "mlx.nn.ALiBi": [[325, null]], "mlx.nn.AvgPool1d": [[326, null]], "mlx.nn.AvgPool2d": [[327, null]], "mlx.nn.BatchNorm": [[328, null]], "mlx.nn.CELU": [[329, null]], "mlx.nn.Conv1d": [[330, null]], "mlx.nn.Conv2d": [[331, null]], "mlx.nn.Conv3d": [[332, null]], "mlx.nn.ConvTranspose1d": [[333, null]], "mlx.nn.ConvTranspose2d": [[334, null]], "mlx.nn.ConvTranspose3d": [[335, null]], "mlx.nn.Dropout": [[336, null]], "mlx.nn.Dropout2d": [[337, null]], "mlx.nn.Dropout3d": [[338, null]], "mlx.nn.ELU": [[339, null]], "mlx.nn.Embedding": [[340, null]], "mlx.nn.GELU": [[341, null]], "mlx.nn.GLU": [[342, null]], "mlx.nn.GRU": [[343, null]], "mlx.nn.GroupNorm": [[344, null]], "mlx.nn.HardShrink": [[345, null]], "mlx.nn.HardTanh": [[346, null]], "mlx.nn.Hardswish": [[347, null]], "mlx.nn.InstanceNorm": [[348, null]], "mlx.nn.LSTM": [[349, null]], "mlx.nn.LayerNorm": [[350, null]], "mlx.nn.LeakyReLU": [[351, null]], "mlx.nn.Linear": [[352, null]], "mlx.nn.LogSigmoid": [[353, null]], "mlx.nn.LogSoftmax": [[354, null]], "mlx.nn.MaxPool1d": [[355, null]], "mlx.nn.MaxPool2d": [[356, null]], "mlx.nn.Mish": [[357, null]], "mlx.nn.Module.apply": [[358, null]], "mlx.nn.Module.apply_to_modules": [[359, null]], "mlx.nn.Module.children": [[360, null]], "mlx.nn.Module.eval": [[361, null]], "mlx.nn.Module.filter_and_map": [[362, null]], "mlx.nn.Module.freeze": [[363, null]], "mlx.nn.Module.leaf_modules": [[364, null]], "mlx.nn.Module.load_weights": [[365, null]], "mlx.nn.Module.modules": [[366, null]], "mlx.nn.Module.named_modules": [[367, null]], "mlx.nn.Module.parameters": [[368, null]], "mlx.nn.Module.save_weights": [[369, null]], "mlx.nn.Module.set_dtype": [[370, null]], "mlx.nn.Module.state": [[371, null]], "mlx.nn.Module.train": [[372, null]], "mlx.nn.Module.trainable_parameters": [[373, null]], "mlx.nn.Module.training": [[374, null]], "mlx.nn.Module.unfreeze": [[375, null]], "mlx.nn.Module.update": [[376, null]], "mlx.nn.Module.update_modules": [[377, null]], "mlx.nn.MultiHeadAttention": [[378, null]], "mlx.nn.PReLU": [[379, null]], "mlx.nn.QuantizedEmbedding": [[380, null]], "mlx.nn.QuantizedLinear": [[381, null]], "mlx.nn.RMSNorm": [[382, null]], "mlx.nn.RNN": [[383, null]], "mlx.nn.ReLU": [[384, null]], "mlx.nn.ReLU6": [[385, null]], "mlx.nn.RoPE": [[386, null]], "mlx.nn.SELU": [[387, null]], "mlx.nn.Sequential": [[388, null]], "mlx.nn.SiLU": [[389, null]], "mlx.nn.Sigmoid": [[390, null]], "mlx.nn.SinusoidalPositionalEncoding": [[391, null]], "mlx.nn.Softmax": [[392, null]], "mlx.nn.Softmin": [[393, null]], "mlx.nn.Softplus": [[394, null]], "mlx.nn.Softshrink": [[395, null]], "mlx.nn.Softsign": [[396, null]], "mlx.nn.Step": [[397, null]], "mlx.nn.Tanh": [[398, null]], "mlx.nn.Transformer": [[399, null]], "mlx.nn.Upsample": [[400, null]], "mlx.nn.celu": [[409, null]], "mlx.nn.elu": [[410, null]], "mlx.nn.gelu": [[411, null]], "mlx.nn.gelu_approx": [[412, null]], "mlx.nn.gelu_fast_approx": [[413, null]], "mlx.nn.glu": [[414, null]], "mlx.nn.hard_shrink": [[415, null]], "mlx.nn.hard_tanh": [[416, null]], "mlx.nn.hardswish": [[417, null]], "mlx.nn.init.constant": [[401, null]], "mlx.nn.init.glorot_normal": [[402, null]], "mlx.nn.init.glorot_uniform": [[403, null]], "mlx.nn.init.he_normal": [[404, null]], "mlx.nn.init.he_uniform": [[405, null]], "mlx.nn.init.identity": [[406, null]], "mlx.nn.init.normal": [[407, null]], "mlx.nn.init.uniform": [[408, null]], "mlx.nn.leaky_relu": [[418, null]], "mlx.nn.log_sigmoid": [[419, null]], "mlx.nn.log_softmax": [[420, null]], "mlx.nn.losses.binary_cross_entropy": [[421, null]], "mlx.nn.losses.cosine_similarity_loss": [[422, null]], "mlx.nn.losses.cross_entropy": [[423, null]], "mlx.nn.losses.gaussian_nll_loss": [[424, null]], "mlx.nn.losses.hinge_loss": [[425, null]], "mlx.nn.losses.huber_loss": [[426, null]], "mlx.nn.losses.kl_div_loss": [[427, null]], "mlx.nn.losses.l1_loss": [[428, null]], "mlx.nn.losses.log_cosh_loss": [[429, null]], "mlx.nn.losses.margin_ranking_loss": [[430, null]], "mlx.nn.losses.mse_loss": [[431, null]], "mlx.nn.losses.nll_loss": [[432, null]], "mlx.nn.losses.smooth_l1_loss": [[433, null]], "mlx.nn.losses.triplet_loss": [[434, null]], "mlx.nn.mish": [[435, null]], "mlx.nn.prelu": [[436, null]], "mlx.nn.quantize": [[307, null]], "mlx.nn.relu": [[437, null]], "mlx.nn.relu6": [[438, null]], "mlx.nn.selu": [[439, null]], "mlx.nn.sigmoid": [[440, null]], "mlx.nn.silu": [[441, null]], "mlx.nn.softmax": [[442, null]], "mlx.nn.softmin": [[443, null]], "mlx.nn.softplus": [[444, null]], "mlx.nn.softshrink": [[445, null]], "mlx.nn.step": [[446, null]], "mlx.nn.tanh": [[447, null]], "mlx.nn.value_and_grad": [[308, null]], "mlx.optimizers.AdaDelta": [[455, null]], "mlx.optimizers.Adafactor": [[456, null]], "mlx.optimizers.Adagrad": [[457, null]], "mlx.optimizers.Adam": [[458, null]], "mlx.optimizers.AdamW": [[459, null]], "mlx.optimizers.Adamax": [[460, null]], "mlx.optimizers.Lion": [[461, null]], "mlx.optimizers.Optimizer.apply_gradients": [[462, null]], "mlx.optimizers.Optimizer.init": [[463, null]], "mlx.optimizers.Optimizer.state": [[464, null]], "mlx.optimizers.Optimizer.update": [[465, null]], "mlx.optimizers.RMSprop": [[466, null]], "mlx.optimizers.SGD": [[467, null]], "mlx.optimizers.clip_grad_norm": [[309, null]], "mlx.optimizers.cosine_decay": [[468, null]], "mlx.optimizers.exponential_decay": [[469, null]], "mlx.optimizers.join_schedules": [[470, null]], "mlx.optimizers.linear_schedule": [[471, null]], "mlx.optimizers.step_decay": [[472, null]], "mlx.utils.tree_flatten": [[310, null]], "mlx.utils.tree_map": [[311, null]], "mlx.utils.tree_map_with_path": [[312, null]], "mlx.utils.tree_reduce": [[313, null]], "mlx.utils.tree_unflatten": [[314, null]], "x86 Shell": [[8, "x86-shell"]]}, "docnames": ["cpp/ops", "dev/custom_metal_kernels", "dev/extensions", "dev/metal_debugger", "examples/linear_regression", "examples/llama-inference", "examples/mlp", "index", "install", "python/_autosummary/mlx.core.Device", "python/_autosummary/mlx.core.Dtype", "python/_autosummary/mlx.core.DtypeCategory", "python/_autosummary/mlx.core.abs", "python/_autosummary/mlx.core.add", "python/_autosummary/mlx.core.addmm", "python/_autosummary/mlx.core.all", "python/_autosummary/mlx.core.allclose", "python/_autosummary/mlx.core.any", "python/_autosummary/mlx.core.arange", "python/_autosummary/mlx.core.arccos", "python/_autosummary/mlx.core.arccosh", "python/_autosummary/mlx.core.arcsin", "python/_autosummary/mlx.core.arcsinh", "python/_autosummary/mlx.core.arctan", "python/_autosummary/mlx.core.arctan2", "python/_autosummary/mlx.core.arctanh", "python/_autosummary/mlx.core.argmax", "python/_autosummary/mlx.core.argmin", "python/_autosummary/mlx.core.argpartition", "python/_autosummary/mlx.core.argsort", "python/_autosummary/mlx.core.array", "python/_autosummary/mlx.core.array.T", "python/_autosummary/mlx.core.array.abs", "python/_autosummary/mlx.core.array.all", "python/_autosummary/mlx.core.array.any", "python/_autosummary/mlx.core.array.argmax", "python/_autosummary/mlx.core.array.argmin", "python/_autosummary/mlx.core.array.astype", "python/_autosummary/mlx.core.array.at", "python/_autosummary/mlx.core.array.conj", "python/_autosummary/mlx.core.array.cos", "python/_autosummary/mlx.core.array.cummax", "python/_autosummary/mlx.core.array.cummin", "python/_autosummary/mlx.core.array.cumprod", "python/_autosummary/mlx.core.array.cumsum", "python/_autosummary/mlx.core.array.diag", "python/_autosummary/mlx.core.array.diagonal", "python/_autosummary/mlx.core.array.dtype", "python/_autosummary/mlx.core.array.exp", "python/_autosummary/mlx.core.array.flatten", "python/_autosummary/mlx.core.array.item", "python/_autosummary/mlx.core.array.itemsize", "python/_autosummary/mlx.core.array.log", "python/_autosummary/mlx.core.array.log10", "python/_autosummary/mlx.core.array.log1p", "python/_autosummary/mlx.core.array.log2", "python/_autosummary/mlx.core.array.logsumexp", "python/_autosummary/mlx.core.array.max", "python/_autosummary/mlx.core.array.mean", "python/_autosummary/mlx.core.array.min", "python/_autosummary/mlx.core.array.moveaxis", "python/_autosummary/mlx.core.array.nbytes", "python/_autosummary/mlx.core.array.ndim", "python/_autosummary/mlx.core.array.prod", "python/_autosummary/mlx.core.array.reciprocal", "python/_autosummary/mlx.core.array.reshape", "python/_autosummary/mlx.core.array.round", "python/_autosummary/mlx.core.array.rsqrt", "python/_autosummary/mlx.core.array.shape", "python/_autosummary/mlx.core.array.sin", "python/_autosummary/mlx.core.array.size", "python/_autosummary/mlx.core.array.split", "python/_autosummary/mlx.core.array.sqrt", "python/_autosummary/mlx.core.array.square", "python/_autosummary/mlx.core.array.squeeze", "python/_autosummary/mlx.core.array.std", "python/_autosummary/mlx.core.array.sum", "python/_autosummary/mlx.core.array.swapaxes", "python/_autosummary/mlx.core.array.tolist", "python/_autosummary/mlx.core.array.transpose", "python/_autosummary/mlx.core.array.var", "python/_autosummary/mlx.core.array.view", "python/_autosummary/mlx.core.array_equal", "python/_autosummary/mlx.core.as_strided", "python/_autosummary/mlx.core.atleast_1d", "python/_autosummary/mlx.core.atleast_2d", "python/_autosummary/mlx.core.atleast_3d", "python/_autosummary/mlx.core.bitwise_and", "python/_autosummary/mlx.core.bitwise_or", "python/_autosummary/mlx.core.bitwise_xor", "python/_autosummary/mlx.core.block_masked_mm", "python/_autosummary/mlx.core.broadcast_to", "python/_autosummary/mlx.core.ceil", "python/_autosummary/mlx.core.clip", "python/_autosummary/mlx.core.compile", "python/_autosummary/mlx.core.concatenate", "python/_autosummary/mlx.core.conj", "python/_autosummary/mlx.core.conjugate", "python/_autosummary/mlx.core.conv1d", "python/_autosummary/mlx.core.conv2d", "python/_autosummary/mlx.core.conv3d", "python/_autosummary/mlx.core.conv_general", "python/_autosummary/mlx.core.conv_transpose1d", "python/_autosummary/mlx.core.conv_transpose2d", "python/_autosummary/mlx.core.conv_transpose3d", "python/_autosummary/mlx.core.convolve", "python/_autosummary/mlx.core.cos", "python/_autosummary/mlx.core.cosh", "python/_autosummary/mlx.core.cummax", "python/_autosummary/mlx.core.cummin", "python/_autosummary/mlx.core.cumprod", "python/_autosummary/mlx.core.cumsum", "python/_autosummary/mlx.core.custom_function", "python/_autosummary/mlx.core.default_device", "python/_autosummary/mlx.core.default_stream", "python/_autosummary/mlx.core.degrees", "python/_autosummary/mlx.core.dequantize", "python/_autosummary/mlx.core.diag", "python/_autosummary/mlx.core.diagonal", "python/_autosummary/mlx.core.disable_compile", "python/_autosummary/mlx.core.distributed.Group", "python/_autosummary/mlx.core.distributed.all_gather", "python/_autosummary/mlx.core.distributed.all_sum", "python/_autosummary/mlx.core.distributed.init", "python/_autosummary/mlx.core.distributed.is_available", "python/_autosummary/mlx.core.distributed.recv", "python/_autosummary/mlx.core.distributed.recv_like", "python/_autosummary/mlx.core.distributed.send", "python/_autosummary/mlx.core.divide", "python/_autosummary/mlx.core.divmod", "python/_autosummary/mlx.core.einsum", "python/_autosummary/mlx.core.einsum_path", "python/_autosummary/mlx.core.enable_compile", "python/_autosummary/mlx.core.equal", "python/_autosummary/mlx.core.erf", "python/_autosummary/mlx.core.erfinv", "python/_autosummary/mlx.core.eval", "python/_autosummary/mlx.core.exp", "python/_autosummary/mlx.core.expand_dims", "python/_autosummary/mlx.core.expm1", "python/_autosummary/mlx.core.eye", "python/_autosummary/mlx.core.fast.affine_quantize", "python/_autosummary/mlx.core.fast.layer_norm", "python/_autosummary/mlx.core.fast.metal_kernel", "python/_autosummary/mlx.core.fast.rms_norm", "python/_autosummary/mlx.core.fast.rope", "python/_autosummary/mlx.core.fast.scaled_dot_product_attention", "python/_autosummary/mlx.core.fft.fft", "python/_autosummary/mlx.core.fft.fft2", "python/_autosummary/mlx.core.fft.fftn", "python/_autosummary/mlx.core.fft.ifft", "python/_autosummary/mlx.core.fft.ifft2", "python/_autosummary/mlx.core.fft.ifftn", "python/_autosummary/mlx.core.fft.irfft", "python/_autosummary/mlx.core.fft.irfft2", "python/_autosummary/mlx.core.fft.irfftn", "python/_autosummary/mlx.core.fft.rfft", "python/_autosummary/mlx.core.fft.rfft2", "python/_autosummary/mlx.core.fft.rfftn", "python/_autosummary/mlx.core.flatten", "python/_autosummary/mlx.core.floor", "python/_autosummary/mlx.core.floor_divide", "python/_autosummary/mlx.core.full", "python/_autosummary/mlx.core.gather_mm", "python/_autosummary/mlx.core.gather_qmm", "python/_autosummary/mlx.core.grad", "python/_autosummary/mlx.core.greater", "python/_autosummary/mlx.core.greater_equal", "python/_autosummary/mlx.core.hadamard_transform", "python/_autosummary/mlx.core.identity", "python/_autosummary/mlx.core.imag", "python/_autosummary/mlx.core.inner", "python/_autosummary/mlx.core.isclose", "python/_autosummary/mlx.core.isfinite", "python/_autosummary/mlx.core.isinf", "python/_autosummary/mlx.core.isnan", "python/_autosummary/mlx.core.isneginf", "python/_autosummary/mlx.core.isposinf", "python/_autosummary/mlx.core.issubdtype", "python/_autosummary/mlx.core.jvp", "python/_autosummary/mlx.core.left_shift", "python/_autosummary/mlx.core.less", "python/_autosummary/mlx.core.less_equal", "python/_autosummary/mlx.core.linalg.cholesky", "python/_autosummary/mlx.core.linalg.cholesky_inv", "python/_autosummary/mlx.core.linalg.cross", "python/_autosummary/mlx.core.linalg.eigh", "python/_autosummary/mlx.core.linalg.eigvalsh", "python/_autosummary/mlx.core.linalg.inv", "python/_autosummary/mlx.core.linalg.norm", "python/_autosummary/mlx.core.linalg.qr", "python/_autosummary/mlx.core.linalg.svd", "python/_autosummary/mlx.core.linalg.tri_inv", "python/_autosummary/mlx.core.linspace", "python/_autosummary/mlx.core.load", "python/_autosummary/mlx.core.log", "python/_autosummary/mlx.core.log10", "python/_autosummary/mlx.core.log1p", "python/_autosummary/mlx.core.log2", "python/_autosummary/mlx.core.logaddexp", "python/_autosummary/mlx.core.logical_and", "python/_autosummary/mlx.core.logical_not", "python/_autosummary/mlx.core.logical_or", "python/_autosummary/mlx.core.logsumexp", "python/_autosummary/mlx.core.matmul", "python/_autosummary/mlx.core.max", "python/_autosummary/mlx.core.maximum", "python/_autosummary/mlx.core.mean", "python/_autosummary/mlx.core.meshgrid", "python/_autosummary/mlx.core.metal.clear_cache", "python/_autosummary/mlx.core.metal.device_info", "python/_autosummary/mlx.core.metal.get_active_memory", "python/_autosummary/mlx.core.metal.get_cache_memory", "python/_autosummary/mlx.core.metal.get_peak_memory", "python/_autosummary/mlx.core.metal.is_available", "python/_autosummary/mlx.core.metal.reset_peak_memory", "python/_autosummary/mlx.core.metal.set_cache_limit", "python/_autosummary/mlx.core.metal.set_memory_limit", "python/_autosummary/mlx.core.metal.set_wired_limit", "python/_autosummary/mlx.core.metal.start_capture", "python/_autosummary/mlx.core.metal.stop_capture", "python/_autosummary/mlx.core.min", "python/_autosummary/mlx.core.minimum", "python/_autosummary/mlx.core.moveaxis", "python/_autosummary/mlx.core.multiply", "python/_autosummary/mlx.core.nan_to_num", "python/_autosummary/mlx.core.negative", "python/_autosummary/mlx.core.new_stream", "python/_autosummary/mlx.core.not_equal", "python/_autosummary/mlx.core.ones", "python/_autosummary/mlx.core.ones_like", "python/_autosummary/mlx.core.outer", "python/_autosummary/mlx.core.pad", "python/_autosummary/mlx.core.partition", "python/_autosummary/mlx.core.power", "python/_autosummary/mlx.core.prod", "python/_autosummary/mlx.core.put_along_axis", "python/_autosummary/mlx.core.quantize", "python/_autosummary/mlx.core.quantized_matmul", "python/_autosummary/mlx.core.radians", "python/_autosummary/mlx.core.random.bernoulli", "python/_autosummary/mlx.core.random.categorical", "python/_autosummary/mlx.core.random.gumbel", "python/_autosummary/mlx.core.random.key", "python/_autosummary/mlx.core.random.laplace", "python/_autosummary/mlx.core.random.multivariate_normal", "python/_autosummary/mlx.core.random.normal", "python/_autosummary/mlx.core.random.permutation", "python/_autosummary/mlx.core.random.randint", "python/_autosummary/mlx.core.random.seed", "python/_autosummary/mlx.core.random.split", "python/_autosummary/mlx.core.random.truncated_normal", "python/_autosummary/mlx.core.random.uniform", "python/_autosummary/mlx.core.real", "python/_autosummary/mlx.core.reciprocal", "python/_autosummary/mlx.core.remainder", "python/_autosummary/mlx.core.repeat", "python/_autosummary/mlx.core.reshape", "python/_autosummary/mlx.core.right_shift", "python/_autosummary/mlx.core.roll", "python/_autosummary/mlx.core.round", "python/_autosummary/mlx.core.rsqrt", "python/_autosummary/mlx.core.save", "python/_autosummary/mlx.core.save_gguf", "python/_autosummary/mlx.core.save_safetensors", "python/_autosummary/mlx.core.savez", "python/_autosummary/mlx.core.savez_compressed", "python/_autosummary/mlx.core.set_default_device", "python/_autosummary/mlx.core.set_default_stream", "python/_autosummary/mlx.core.sigmoid", "python/_autosummary/mlx.core.sign", "python/_autosummary/mlx.core.sin", "python/_autosummary/mlx.core.sinh", "python/_autosummary/mlx.core.softmax", "python/_autosummary/mlx.core.sort", "python/_autosummary/mlx.core.split", "python/_autosummary/mlx.core.sqrt", "python/_autosummary/mlx.core.square", "python/_autosummary/mlx.core.squeeze", "python/_autosummary/mlx.core.stack", "python/_autosummary/mlx.core.std", "python/_autosummary/mlx.core.stop_gradient", "python/_autosummary/mlx.core.stream", "python/_autosummary/mlx.core.subtract", "python/_autosummary/mlx.core.sum", "python/_autosummary/mlx.core.swapaxes", "python/_autosummary/mlx.core.synchronize", "python/_autosummary/mlx.core.take", "python/_autosummary/mlx.core.take_along_axis", "python/_autosummary/mlx.core.tan", "python/_autosummary/mlx.core.tanh", "python/_autosummary/mlx.core.tensordot", "python/_autosummary/mlx.core.tile", "python/_autosummary/mlx.core.topk", "python/_autosummary/mlx.core.trace", "python/_autosummary/mlx.core.transpose", "python/_autosummary/mlx.core.tri", "python/_autosummary/mlx.core.tril", "python/_autosummary/mlx.core.triu", "python/_autosummary/mlx.core.value_and_grad", "python/_autosummary/mlx.core.var", "python/_autosummary/mlx.core.view", "python/_autosummary/mlx.core.vjp", "python/_autosummary/mlx.core.vmap", "python/_autosummary/mlx.core.where", "python/_autosummary/mlx.core.zeros", "python/_autosummary/mlx.core.zeros_like", "python/_autosummary/mlx.nn.quantize", "python/_autosummary/mlx.nn.value_and_grad", "python/_autosummary/mlx.optimizers.clip_grad_norm", "python/_autosummary/mlx.utils.tree_flatten", "python/_autosummary/mlx.utils.tree_map", "python/_autosummary/mlx.utils.tree_map_with_path", "python/_autosummary/mlx.utils.tree_reduce", "python/_autosummary/mlx.utils.tree_unflatten", "python/_autosummary/stream_class", "python/array", "python/data_types", "python/devices_and_streams", "python/distributed", "python/fast", "python/fft", "python/linalg", "python/metal", "python/nn", "python/nn/_autosummary/mlx.nn.ALiBi", "python/nn/_autosummary/mlx.nn.AvgPool1d", "python/nn/_autosummary/mlx.nn.AvgPool2d", "python/nn/_autosummary/mlx.nn.BatchNorm", "python/nn/_autosummary/mlx.nn.CELU", "python/nn/_autosummary/mlx.nn.Conv1d", "python/nn/_autosummary/mlx.nn.Conv2d", "python/nn/_autosummary/mlx.nn.Conv3d", "python/nn/_autosummary/mlx.nn.ConvTranspose1d", "python/nn/_autosummary/mlx.nn.ConvTranspose2d", "python/nn/_autosummary/mlx.nn.ConvTranspose3d", "python/nn/_autosummary/mlx.nn.Dropout", "python/nn/_autosummary/mlx.nn.Dropout2d", "python/nn/_autosummary/mlx.nn.Dropout3d", "python/nn/_autosummary/mlx.nn.ELU", "python/nn/_autosummary/mlx.nn.Embedding", "python/nn/_autosummary/mlx.nn.GELU", "python/nn/_autosummary/mlx.nn.GLU", "python/nn/_autosummary/mlx.nn.GRU", "python/nn/_autosummary/mlx.nn.GroupNorm", "python/nn/_autosummary/mlx.nn.HardShrink", "python/nn/_autosummary/mlx.nn.HardTanh", "python/nn/_autosummary/mlx.nn.Hardswish", "python/nn/_autosummary/mlx.nn.InstanceNorm", "python/nn/_autosummary/mlx.nn.LSTM", "python/nn/_autosummary/mlx.nn.LayerNorm", "python/nn/_autosummary/mlx.nn.LeakyReLU", "python/nn/_autosummary/mlx.nn.Linear", "python/nn/_autosummary/mlx.nn.LogSigmoid", "python/nn/_autosummary/mlx.nn.LogSoftmax", "python/nn/_autosummary/mlx.nn.MaxPool1d", "python/nn/_autosummary/mlx.nn.MaxPool2d", "python/nn/_autosummary/mlx.nn.Mish", "python/nn/_autosummary/mlx.nn.Module.apply", "python/nn/_autosummary/mlx.nn.Module.apply_to_modules", "python/nn/_autosummary/mlx.nn.Module.children", "python/nn/_autosummary/mlx.nn.Module.eval", "python/nn/_autosummary/mlx.nn.Module.filter_and_map", "python/nn/_autosummary/mlx.nn.Module.freeze", "python/nn/_autosummary/mlx.nn.Module.leaf_modules", "python/nn/_autosummary/mlx.nn.Module.load_weights", "python/nn/_autosummary/mlx.nn.Module.modules", "python/nn/_autosummary/mlx.nn.Module.named_modules", "python/nn/_autosummary/mlx.nn.Module.parameters", "python/nn/_autosummary/mlx.nn.Module.save_weights", "python/nn/_autosummary/mlx.nn.Module.set_dtype", "python/nn/_autosummary/mlx.nn.Module.state", "python/nn/_autosummary/mlx.nn.Module.train", "python/nn/_autosummary/mlx.nn.Module.trainable_parameters", "python/nn/_autosummary/mlx.nn.Module.training", "python/nn/_autosummary/mlx.nn.Module.unfreeze", "python/nn/_autosummary/mlx.nn.Module.update", "python/nn/_autosummary/mlx.nn.Module.update_modules", "python/nn/_autosummary/mlx.nn.MultiHeadAttention", "python/nn/_autosummary/mlx.nn.PReLU", "python/nn/_autosummary/mlx.nn.QuantizedEmbedding", "python/nn/_autosummary/mlx.nn.QuantizedLinear", "python/nn/_autosummary/mlx.nn.RMSNorm", "python/nn/_autosummary/mlx.nn.RNN", "python/nn/_autosummary/mlx.nn.ReLU", "python/nn/_autosummary/mlx.nn.ReLU6", "python/nn/_autosummary/mlx.nn.RoPE", "python/nn/_autosummary/mlx.nn.SELU", "python/nn/_autosummary/mlx.nn.Sequential", "python/nn/_autosummary/mlx.nn.SiLU", "python/nn/_autosummary/mlx.nn.Sigmoid", "python/nn/_autosummary/mlx.nn.SinusoidalPositionalEncoding", "python/nn/_autosummary/mlx.nn.Softmax", "python/nn/_autosummary/mlx.nn.Softmin", "python/nn/_autosummary/mlx.nn.Softplus", "python/nn/_autosummary/mlx.nn.Softshrink", "python/nn/_autosummary/mlx.nn.Softsign", "python/nn/_autosummary/mlx.nn.Step", "python/nn/_autosummary/mlx.nn.Tanh", "python/nn/_autosummary/mlx.nn.Transformer", "python/nn/_autosummary/mlx.nn.Upsample", "python/nn/_autosummary/mlx.nn.init.constant", "python/nn/_autosummary/mlx.nn.init.glorot_normal", "python/nn/_autosummary/mlx.nn.init.glorot_uniform", "python/nn/_autosummary/mlx.nn.init.he_normal", "python/nn/_autosummary/mlx.nn.init.he_uniform", "python/nn/_autosummary/mlx.nn.init.identity", "python/nn/_autosummary/mlx.nn.init.normal", "python/nn/_autosummary/mlx.nn.init.uniform", "python/nn/_autosummary_functions/mlx.nn.celu", "python/nn/_autosummary_functions/mlx.nn.elu", "python/nn/_autosummary_functions/mlx.nn.gelu", "python/nn/_autosummary_functions/mlx.nn.gelu_approx", "python/nn/_autosummary_functions/mlx.nn.gelu_fast_approx", "python/nn/_autosummary_functions/mlx.nn.glu", "python/nn/_autosummary_functions/mlx.nn.hard_shrink", "python/nn/_autosummary_functions/mlx.nn.hard_tanh", "python/nn/_autosummary_functions/mlx.nn.hardswish", "python/nn/_autosummary_functions/mlx.nn.leaky_relu", "python/nn/_autosummary_functions/mlx.nn.log_sigmoid", "python/nn/_autosummary_functions/mlx.nn.log_softmax", "python/nn/_autosummary_functions/mlx.nn.losses.binary_cross_entropy", "python/nn/_autosummary_functions/mlx.nn.losses.cosine_similarity_loss", "python/nn/_autosummary_functions/mlx.nn.losses.cross_entropy", "python/nn/_autosummary_functions/mlx.nn.losses.gaussian_nll_loss", "python/nn/_autosummary_functions/mlx.nn.losses.hinge_loss", "python/nn/_autosummary_functions/mlx.nn.losses.huber_loss", "python/nn/_autosummary_functions/mlx.nn.losses.kl_div_loss", "python/nn/_autosummary_functions/mlx.nn.losses.l1_loss", "python/nn/_autosummary_functions/mlx.nn.losses.log_cosh_loss", "python/nn/_autosummary_functions/mlx.nn.losses.margin_ranking_loss", "python/nn/_autosummary_functions/mlx.nn.losses.mse_loss", "python/nn/_autosummary_functions/mlx.nn.losses.nll_loss", "python/nn/_autosummary_functions/mlx.nn.losses.smooth_l1_loss", "python/nn/_autosummary_functions/mlx.nn.losses.triplet_loss", "python/nn/_autosummary_functions/mlx.nn.mish", "python/nn/_autosummary_functions/mlx.nn.prelu", "python/nn/_autosummary_functions/mlx.nn.relu", "python/nn/_autosummary_functions/mlx.nn.relu6", "python/nn/_autosummary_functions/mlx.nn.selu", "python/nn/_autosummary_functions/mlx.nn.sigmoid", "python/nn/_autosummary_functions/mlx.nn.silu", "python/nn/_autosummary_functions/mlx.nn.softmax", "python/nn/_autosummary_functions/mlx.nn.softmin", "python/nn/_autosummary_functions/mlx.nn.softplus", "python/nn/_autosummary_functions/mlx.nn.softshrink", "python/nn/_autosummary_functions/mlx.nn.step", "python/nn/_autosummary_functions/mlx.nn.tanh", "python/nn/functions", "python/nn/init", "python/nn/layers", "python/nn/losses", "python/nn/module", "python/ops", "python/optimizers", "python/optimizers/_autosummary/mlx.optimizers.AdaDelta", "python/optimizers/_autosummary/mlx.optimizers.Adafactor", "python/optimizers/_autosummary/mlx.optimizers.Adagrad", "python/optimizers/_autosummary/mlx.optimizers.Adam", "python/optimizers/_autosummary/mlx.optimizers.AdamW", "python/optimizers/_autosummary/mlx.optimizers.Adamax", "python/optimizers/_autosummary/mlx.optimizers.Lion", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.apply_gradients", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.init", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.state", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.update", "python/optimizers/_autosummary/mlx.optimizers.RMSprop", "python/optimizers/_autosummary/mlx.optimizers.SGD", "python/optimizers/_autosummary/mlx.optimizers.cosine_decay", "python/optimizers/_autosummary/mlx.optimizers.exponential_decay", "python/optimizers/_autosummary/mlx.optimizers.join_schedules", "python/optimizers/_autosummary/mlx.optimizers.linear_schedule", "python/optimizers/_autosummary/mlx.optimizers.step_decay", "python/optimizers/common_optimizers", "python/optimizers/optimizer", "python/optimizers/schedulers", "python/random", "python/transforms", "python/tree_utils", "usage/compile", "usage/distributed", "usage/function_transforms", "usage/indexing", "usage/lazy_evaluation", "usage/numpy", "usage/quick_start", "usage/saving_and_loading", "usage/unified_memory", "usage/using_streams"], "envversion": {"sphinx": 62, "sphinx.domains.c": 3, "sphinx.domains.changeset": 1, "sphinx.domains.citation": 1, "sphinx.domains.cpp": 9, "sphinx.domains.index": 1, "sphinx.domains.javascript": 3, "sphinx.domains.math": 2, "sphinx.domains.python": 4, "sphinx.domains.rst": 2, "sphinx.domains.std": 2, "sphinx.ext.intersphinx": 1}, "filenames": ["cpp/ops.rst", "dev/custom_metal_kernels.rst", "dev/extensions.rst", "dev/metal_debugger.rst", "examples/linear_regression.rst", "examples/llama-inference.rst", "examples/mlp.rst", "index.rst", "install.rst", "python/_autosummary/mlx.core.Device.rst", "python/_autosummary/mlx.core.Dtype.rst", "python/_autosummary/mlx.core.DtypeCategory.rst", "python/_autosummary/mlx.core.abs.rst", "python/_autosummary/mlx.core.add.rst", "python/_autosummary/mlx.core.addmm.rst", "python/_autosummary/mlx.core.all.rst", "python/_autosummary/mlx.core.allclose.rst", "python/_autosummary/mlx.core.any.rst", "python/_autosummary/mlx.core.arange.rst", "python/_autosummary/mlx.core.arccos.rst", "python/_autosummary/mlx.core.arccosh.rst", "python/_autosummary/mlx.core.arcsin.rst", "python/_autosummary/mlx.core.arcsinh.rst", "python/_autosummary/mlx.core.arctan.rst", "python/_autosummary/mlx.core.arctan2.rst", "python/_autosummary/mlx.core.arctanh.rst", "python/_autosummary/mlx.core.argmax.rst", "python/_autosummary/mlx.core.argmin.rst", "python/_autosummary/mlx.core.argpartition.rst", "python/_autosummary/mlx.core.argsort.rst", "python/_autosummary/mlx.core.array.rst", "python/_autosummary/mlx.core.array.T.rst", "python/_autosummary/mlx.core.array.abs.rst", "python/_autosummary/mlx.core.array.all.rst", "python/_autosummary/mlx.core.array.any.rst", "python/_autosummary/mlx.core.array.argmax.rst", "python/_autosummary/mlx.core.array.argmin.rst", "python/_autosummary/mlx.core.array.astype.rst", "python/_autosummary/mlx.core.array.at.rst", "python/_autosummary/mlx.core.array.conj.rst", "python/_autosummary/mlx.core.array.cos.rst", "python/_autosummary/mlx.core.array.cummax.rst", "python/_autosummary/mlx.core.array.cummin.rst", "python/_autosummary/mlx.core.array.cumprod.rst", "python/_autosummary/mlx.core.array.cumsum.rst", "python/_autosummary/mlx.core.array.diag.rst", "python/_autosummary/mlx.core.array.diagonal.rst", "python/_autosummary/mlx.core.array.dtype.rst", "python/_autosummary/mlx.core.array.exp.rst", "python/_autosummary/mlx.core.array.flatten.rst", "python/_autosummary/mlx.core.array.item.rst", "python/_autosummary/mlx.core.array.itemsize.rst", "python/_autosummary/mlx.core.array.log.rst", "python/_autosummary/mlx.core.array.log10.rst", "python/_autosummary/mlx.core.array.log1p.rst", "python/_autosummary/mlx.core.array.log2.rst", "python/_autosummary/mlx.core.array.logsumexp.rst", "python/_autosummary/mlx.core.array.max.rst", "python/_autosummary/mlx.core.array.mean.rst", "python/_autosummary/mlx.core.array.min.rst", "python/_autosummary/mlx.core.array.moveaxis.rst", "python/_autosummary/mlx.core.array.nbytes.rst", "python/_autosummary/mlx.core.array.ndim.rst", "python/_autosummary/mlx.core.array.prod.rst", "python/_autosummary/mlx.core.array.reciprocal.rst", "python/_autosummary/mlx.core.array.reshape.rst", "python/_autosummary/mlx.core.array.round.rst", "python/_autosummary/mlx.core.array.rsqrt.rst", "python/_autosummary/mlx.core.array.shape.rst", "python/_autosummary/mlx.core.array.sin.rst", "python/_autosummary/mlx.core.array.size.rst", "python/_autosummary/mlx.core.array.split.rst", "python/_autosummary/mlx.core.array.sqrt.rst", "python/_autosummary/mlx.core.array.square.rst", "python/_autosummary/mlx.core.array.squeeze.rst", "python/_autosummary/mlx.core.array.std.rst", "python/_autosummary/mlx.core.array.sum.rst", "python/_autosummary/mlx.core.array.swapaxes.rst", "python/_autosummary/mlx.core.array.tolist.rst", "python/_autosummary/mlx.core.array.transpose.rst", "python/_autosummary/mlx.core.array.var.rst", "python/_autosummary/mlx.core.array.view.rst", "python/_autosummary/mlx.core.array_equal.rst", "python/_autosummary/mlx.core.as_strided.rst", "python/_autosummary/mlx.core.atleast_1d.rst", "python/_autosummary/mlx.core.atleast_2d.rst", "python/_autosummary/mlx.core.atleast_3d.rst", "python/_autosummary/mlx.core.bitwise_and.rst", "python/_autosummary/mlx.core.bitwise_or.rst", "python/_autosummary/mlx.core.bitwise_xor.rst", "python/_autosummary/mlx.core.block_masked_mm.rst", "python/_autosummary/mlx.core.broadcast_to.rst", "python/_autosummary/mlx.core.ceil.rst", "python/_autosummary/mlx.core.clip.rst", "python/_autosummary/mlx.core.compile.rst", "python/_autosummary/mlx.core.concatenate.rst", "python/_autosummary/mlx.core.conj.rst", "python/_autosummary/mlx.core.conjugate.rst", "python/_autosummary/mlx.core.conv1d.rst", "python/_autosummary/mlx.core.conv2d.rst", "python/_autosummary/mlx.core.conv3d.rst", "python/_autosummary/mlx.core.conv_general.rst", "python/_autosummary/mlx.core.conv_transpose1d.rst", "python/_autosummary/mlx.core.conv_transpose2d.rst", "python/_autosummary/mlx.core.conv_transpose3d.rst", "python/_autosummary/mlx.core.convolve.rst", "python/_autosummary/mlx.core.cos.rst", "python/_autosummary/mlx.core.cosh.rst", "python/_autosummary/mlx.core.cummax.rst", "python/_autosummary/mlx.core.cummin.rst", "python/_autosummary/mlx.core.cumprod.rst", "python/_autosummary/mlx.core.cumsum.rst", "python/_autosummary/mlx.core.custom_function.rst", "python/_autosummary/mlx.core.default_device.rst", "python/_autosummary/mlx.core.default_stream.rst", "python/_autosummary/mlx.core.degrees.rst", "python/_autosummary/mlx.core.dequantize.rst", "python/_autosummary/mlx.core.diag.rst", "python/_autosummary/mlx.core.diagonal.rst", "python/_autosummary/mlx.core.disable_compile.rst", "python/_autosummary/mlx.core.distributed.Group.rst", "python/_autosummary/mlx.core.distributed.all_gather.rst", "python/_autosummary/mlx.core.distributed.all_sum.rst", "python/_autosummary/mlx.core.distributed.init.rst", "python/_autosummary/mlx.core.distributed.is_available.rst", "python/_autosummary/mlx.core.distributed.recv.rst", "python/_autosummary/mlx.core.distributed.recv_like.rst", "python/_autosummary/mlx.core.distributed.send.rst", "python/_autosummary/mlx.core.divide.rst", "python/_autosummary/mlx.core.divmod.rst", "python/_autosummary/mlx.core.einsum.rst", "python/_autosummary/mlx.core.einsum_path.rst", "python/_autosummary/mlx.core.enable_compile.rst", "python/_autosummary/mlx.core.equal.rst", "python/_autosummary/mlx.core.erf.rst", "python/_autosummary/mlx.core.erfinv.rst", "python/_autosummary/mlx.core.eval.rst", "python/_autosummary/mlx.core.exp.rst", "python/_autosummary/mlx.core.expand_dims.rst", "python/_autosummary/mlx.core.expm1.rst", "python/_autosummary/mlx.core.eye.rst", "python/_autosummary/mlx.core.fast.affine_quantize.rst", "python/_autosummary/mlx.core.fast.layer_norm.rst", "python/_autosummary/mlx.core.fast.metal_kernel.rst", "python/_autosummary/mlx.core.fast.rms_norm.rst", "python/_autosummary/mlx.core.fast.rope.rst", "python/_autosummary/mlx.core.fast.scaled_dot_product_attention.rst", "python/_autosummary/mlx.core.fft.fft.rst", "python/_autosummary/mlx.core.fft.fft2.rst", "python/_autosummary/mlx.core.fft.fftn.rst", "python/_autosummary/mlx.core.fft.ifft.rst", "python/_autosummary/mlx.core.fft.ifft2.rst", "python/_autosummary/mlx.core.fft.ifftn.rst", "python/_autosummary/mlx.core.fft.irfft.rst", "python/_autosummary/mlx.core.fft.irfft2.rst", "python/_autosummary/mlx.core.fft.irfftn.rst", "python/_autosummary/mlx.core.fft.rfft.rst", "python/_autosummary/mlx.core.fft.rfft2.rst", "python/_autosummary/mlx.core.fft.rfftn.rst", "python/_autosummary/mlx.core.flatten.rst", "python/_autosummary/mlx.core.floor.rst", "python/_autosummary/mlx.core.floor_divide.rst", "python/_autosummary/mlx.core.full.rst", "python/_autosummary/mlx.core.gather_mm.rst", "python/_autosummary/mlx.core.gather_qmm.rst", "python/_autosummary/mlx.core.grad.rst", "python/_autosummary/mlx.core.greater.rst", "python/_autosummary/mlx.core.greater_equal.rst", "python/_autosummary/mlx.core.hadamard_transform.rst", "python/_autosummary/mlx.core.identity.rst", "python/_autosummary/mlx.core.imag.rst", "python/_autosummary/mlx.core.inner.rst", "python/_autosummary/mlx.core.isclose.rst", "python/_autosummary/mlx.core.isfinite.rst", "python/_autosummary/mlx.core.isinf.rst", "python/_autosummary/mlx.core.isnan.rst", "python/_autosummary/mlx.core.isneginf.rst", "python/_autosummary/mlx.core.isposinf.rst", "python/_autosummary/mlx.core.issubdtype.rst", "python/_autosummary/mlx.core.jvp.rst", "python/_autosummary/mlx.core.left_shift.rst", "python/_autosummary/mlx.core.less.rst", "python/_autosummary/mlx.core.less_equal.rst", "python/_autosummary/mlx.core.linalg.cholesky.rst", "python/_autosummary/mlx.core.linalg.cholesky_inv.rst", "python/_autosummary/mlx.core.linalg.cross.rst", "python/_autosummary/mlx.core.linalg.eigh.rst", "python/_autosummary/mlx.core.linalg.eigvalsh.rst", "python/_autosummary/mlx.core.linalg.inv.rst", "python/_autosummary/mlx.core.linalg.norm.rst", "python/_autosummary/mlx.core.linalg.qr.rst", "python/_autosummary/mlx.core.linalg.svd.rst", "python/_autosummary/mlx.core.linalg.tri_inv.rst", "python/_autosummary/mlx.core.linspace.rst", "python/_autosummary/mlx.core.load.rst", "python/_autosummary/mlx.core.log.rst", "python/_autosummary/mlx.core.log10.rst", "python/_autosummary/mlx.core.log1p.rst", "python/_autosummary/mlx.core.log2.rst", "python/_autosummary/mlx.core.logaddexp.rst", "python/_autosummary/mlx.core.logical_and.rst", "python/_autosummary/mlx.core.logical_not.rst", "python/_autosummary/mlx.core.logical_or.rst", "python/_autosummary/mlx.core.logsumexp.rst", "python/_autosummary/mlx.core.matmul.rst", "python/_autosummary/mlx.core.max.rst", "python/_autosummary/mlx.core.maximum.rst", "python/_autosummary/mlx.core.mean.rst", "python/_autosummary/mlx.core.meshgrid.rst", "python/_autosummary/mlx.core.metal.clear_cache.rst", "python/_autosummary/mlx.core.metal.device_info.rst", "python/_autosummary/mlx.core.metal.get_active_memory.rst", "python/_autosummary/mlx.core.metal.get_cache_memory.rst", "python/_autosummary/mlx.core.metal.get_peak_memory.rst", "python/_autosummary/mlx.core.metal.is_available.rst", "python/_autosummary/mlx.core.metal.reset_peak_memory.rst", "python/_autosummary/mlx.core.metal.set_cache_limit.rst", "python/_autosummary/mlx.core.metal.set_memory_limit.rst", "python/_autosummary/mlx.core.metal.set_wired_limit.rst", "python/_autosummary/mlx.core.metal.start_capture.rst", "python/_autosummary/mlx.core.metal.stop_capture.rst", "python/_autosummary/mlx.core.min.rst", "python/_autosummary/mlx.core.minimum.rst", "python/_autosummary/mlx.core.moveaxis.rst", "python/_autosummary/mlx.core.multiply.rst", "python/_autosummary/mlx.core.nan_to_num.rst", "python/_autosummary/mlx.core.negative.rst", "python/_autosummary/mlx.core.new_stream.rst", "python/_autosummary/mlx.core.not_equal.rst", "python/_autosummary/mlx.core.ones.rst", "python/_autosummary/mlx.core.ones_like.rst", "python/_autosummary/mlx.core.outer.rst", "python/_autosummary/mlx.core.pad.rst", "python/_autosummary/mlx.core.partition.rst", "python/_autosummary/mlx.core.power.rst", "python/_autosummary/mlx.core.prod.rst", "python/_autosummary/mlx.core.put_along_axis.rst", "python/_autosummary/mlx.core.quantize.rst", "python/_autosummary/mlx.core.quantized_matmul.rst", "python/_autosummary/mlx.core.radians.rst", "python/_autosummary/mlx.core.random.bernoulli.rst", "python/_autosummary/mlx.core.random.categorical.rst", "python/_autosummary/mlx.core.random.gumbel.rst", "python/_autosummary/mlx.core.random.key.rst", "python/_autosummary/mlx.core.random.laplace.rst", "python/_autosummary/mlx.core.random.multivariate_normal.rst", "python/_autosummary/mlx.core.random.normal.rst", "python/_autosummary/mlx.core.random.permutation.rst", "python/_autosummary/mlx.core.random.randint.rst", "python/_autosummary/mlx.core.random.seed.rst", "python/_autosummary/mlx.core.random.split.rst", "python/_autosummary/mlx.core.random.truncated_normal.rst", "python/_autosummary/mlx.core.random.uniform.rst", "python/_autosummary/mlx.core.real.rst", "python/_autosummary/mlx.core.reciprocal.rst", "python/_autosummary/mlx.core.remainder.rst", "python/_autosummary/mlx.core.repeat.rst", "python/_autosummary/mlx.core.reshape.rst", "python/_autosummary/mlx.core.right_shift.rst", "python/_autosummary/mlx.core.roll.rst", "python/_autosummary/mlx.core.round.rst", "python/_autosummary/mlx.core.rsqrt.rst", "python/_autosummary/mlx.core.save.rst", "python/_autosummary/mlx.core.save_gguf.rst", "python/_autosummary/mlx.core.save_safetensors.rst", "python/_autosummary/mlx.core.savez.rst", "python/_autosummary/mlx.core.savez_compressed.rst", "python/_autosummary/mlx.core.set_default_device.rst", "python/_autosummary/mlx.core.set_default_stream.rst", "python/_autosummary/mlx.core.sigmoid.rst", "python/_autosummary/mlx.core.sign.rst", "python/_autosummary/mlx.core.sin.rst", "python/_autosummary/mlx.core.sinh.rst", "python/_autosummary/mlx.core.softmax.rst", "python/_autosummary/mlx.core.sort.rst", "python/_autosummary/mlx.core.split.rst", "python/_autosummary/mlx.core.sqrt.rst", "python/_autosummary/mlx.core.square.rst", "python/_autosummary/mlx.core.squeeze.rst", "python/_autosummary/mlx.core.stack.rst", "python/_autosummary/mlx.core.std.rst", "python/_autosummary/mlx.core.stop_gradient.rst", "python/_autosummary/mlx.core.stream.rst", "python/_autosummary/mlx.core.subtract.rst", "python/_autosummary/mlx.core.sum.rst", "python/_autosummary/mlx.core.swapaxes.rst", "python/_autosummary/mlx.core.synchronize.rst", "python/_autosummary/mlx.core.take.rst", "python/_autosummary/mlx.core.take_along_axis.rst", "python/_autosummary/mlx.core.tan.rst", "python/_autosummary/mlx.core.tanh.rst", "python/_autosummary/mlx.core.tensordot.rst", "python/_autosummary/mlx.core.tile.rst", "python/_autosummary/mlx.core.topk.rst", "python/_autosummary/mlx.core.trace.rst", "python/_autosummary/mlx.core.transpose.rst", "python/_autosummary/mlx.core.tri.rst", "python/_autosummary/mlx.core.tril.rst", "python/_autosummary/mlx.core.triu.rst", "python/_autosummary/mlx.core.value_and_grad.rst", "python/_autosummary/mlx.core.var.rst", "python/_autosummary/mlx.core.view.rst", "python/_autosummary/mlx.core.vjp.rst", "python/_autosummary/mlx.core.vmap.rst", "python/_autosummary/mlx.core.where.rst", "python/_autosummary/mlx.core.zeros.rst", "python/_autosummary/mlx.core.zeros_like.rst", "python/_autosummary/mlx.nn.quantize.rst", "python/_autosummary/mlx.nn.value_and_grad.rst", "python/_autosummary/mlx.optimizers.clip_grad_norm.rst", "python/_autosummary/mlx.utils.tree_flatten.rst", "python/_autosummary/mlx.utils.tree_map.rst", "python/_autosummary/mlx.utils.tree_map_with_path.rst", "python/_autosummary/mlx.utils.tree_reduce.rst", "python/_autosummary/mlx.utils.tree_unflatten.rst", "python/_autosummary/stream_class.rst", "python/array.rst", "python/data_types.rst", "python/devices_and_streams.rst", "python/distributed.rst", "python/fast.rst", "python/fft.rst", "python/linalg.rst", "python/metal.rst", "python/nn.rst", "python/nn/_autosummary/mlx.nn.ALiBi.rst", "python/nn/_autosummary/mlx.nn.AvgPool1d.rst", "python/nn/_autosummary/mlx.nn.AvgPool2d.rst", "python/nn/_autosummary/mlx.nn.BatchNorm.rst", "python/nn/_autosummary/mlx.nn.CELU.rst", "python/nn/_autosummary/mlx.nn.Conv1d.rst", "python/nn/_autosummary/mlx.nn.Conv2d.rst", "python/nn/_autosummary/mlx.nn.Conv3d.rst", "python/nn/_autosummary/mlx.nn.ConvTranspose1d.rst", "python/nn/_autosummary/mlx.nn.ConvTranspose2d.rst", "python/nn/_autosummary/mlx.nn.ConvTranspose3d.rst", "python/nn/_autosummary/mlx.nn.Dropout.rst", "python/nn/_autosummary/mlx.nn.Dropout2d.rst", "python/nn/_autosummary/mlx.nn.Dropout3d.rst", "python/nn/_autosummary/mlx.nn.ELU.rst", "python/nn/_autosummary/mlx.nn.Embedding.rst", "python/nn/_autosummary/mlx.nn.GELU.rst", "python/nn/_autosummary/mlx.nn.GLU.rst", "python/nn/_autosummary/mlx.nn.GRU.rst", "python/nn/_autosummary/mlx.nn.GroupNorm.rst", "python/nn/_autosummary/mlx.nn.HardShrink.rst", "python/nn/_autosummary/mlx.nn.HardTanh.rst", "python/nn/_autosummary/mlx.nn.Hardswish.rst", "python/nn/_autosummary/mlx.nn.InstanceNorm.rst", "python/nn/_autosummary/mlx.nn.LSTM.rst", "python/nn/_autosummary/mlx.nn.LayerNorm.rst", "python/nn/_autosummary/mlx.nn.LeakyReLU.rst", "python/nn/_autosummary/mlx.nn.Linear.rst", "python/nn/_autosummary/mlx.nn.LogSigmoid.rst", "python/nn/_autosummary/mlx.nn.LogSoftmax.rst", "python/nn/_autosummary/mlx.nn.MaxPool1d.rst", "python/nn/_autosummary/mlx.nn.MaxPool2d.rst", "python/nn/_autosummary/mlx.nn.Mish.rst", "python/nn/_autosummary/mlx.nn.Module.apply.rst", "python/nn/_autosummary/mlx.nn.Module.apply_to_modules.rst", "python/nn/_autosummary/mlx.nn.Module.children.rst", "python/nn/_autosummary/mlx.nn.Module.eval.rst", "python/nn/_autosummary/mlx.nn.Module.filter_and_map.rst", "python/nn/_autosummary/mlx.nn.Module.freeze.rst", "python/nn/_autosummary/mlx.nn.Module.leaf_modules.rst", "python/nn/_autosummary/mlx.nn.Module.load_weights.rst", "python/nn/_autosummary/mlx.nn.Module.modules.rst", "python/nn/_autosummary/mlx.nn.Module.named_modules.rst", "python/nn/_autosummary/mlx.nn.Module.parameters.rst", "python/nn/_autosummary/mlx.nn.Module.save_weights.rst", "python/nn/_autosummary/mlx.nn.Module.set_dtype.rst", "python/nn/_autosummary/mlx.nn.Module.state.rst", "python/nn/_autosummary/mlx.nn.Module.train.rst", "python/nn/_autosummary/mlx.nn.Module.trainable_parameters.rst", "python/nn/_autosummary/mlx.nn.Module.training.rst", "python/nn/_autosummary/mlx.nn.Module.unfreeze.rst", "python/nn/_autosummary/mlx.nn.Module.update.rst", "python/nn/_autosummary/mlx.nn.Module.update_modules.rst", "python/nn/_autosummary/mlx.nn.MultiHeadAttention.rst", "python/nn/_autosummary/mlx.nn.PReLU.rst", "python/nn/_autosummary/mlx.nn.QuantizedEmbedding.rst", "python/nn/_autosummary/mlx.nn.QuantizedLinear.rst", "python/nn/_autosummary/mlx.nn.RMSNorm.rst", "python/nn/_autosummary/mlx.nn.RNN.rst", "python/nn/_autosummary/mlx.nn.ReLU.rst", "python/nn/_autosummary/mlx.nn.ReLU6.rst", "python/nn/_autosummary/mlx.nn.RoPE.rst", "python/nn/_autosummary/mlx.nn.SELU.rst", "python/nn/_autosummary/mlx.nn.Sequential.rst", "python/nn/_autosummary/mlx.nn.SiLU.rst", "python/nn/_autosummary/mlx.nn.Sigmoid.rst", "python/nn/_autosummary/mlx.nn.SinusoidalPositionalEncoding.rst", "python/nn/_autosummary/mlx.nn.Softmax.rst", "python/nn/_autosummary/mlx.nn.Softmin.rst", "python/nn/_autosummary/mlx.nn.Softplus.rst", "python/nn/_autosummary/mlx.nn.Softshrink.rst", "python/nn/_autosummary/mlx.nn.Softsign.rst", "python/nn/_autosummary/mlx.nn.Step.rst", "python/nn/_autosummary/mlx.nn.Tanh.rst", "python/nn/_autosummary/mlx.nn.Transformer.rst", "python/nn/_autosummary/mlx.nn.Upsample.rst", "python/nn/_autosummary/mlx.nn.init.constant.rst", "python/nn/_autosummary/mlx.nn.init.glorot_normal.rst", "python/nn/_autosummary/mlx.nn.init.glorot_uniform.rst", "python/nn/_autosummary/mlx.nn.init.he_normal.rst", "python/nn/_autosummary/mlx.nn.init.he_uniform.rst", "python/nn/_autosummary/mlx.nn.init.identity.rst", "python/nn/_autosummary/mlx.nn.init.normal.rst", "python/nn/_autosummary/mlx.nn.init.uniform.rst", "python/nn/_autosummary_functions/mlx.nn.celu.rst", "python/nn/_autosummary_functions/mlx.nn.elu.rst", "python/nn/_autosummary_functions/mlx.nn.gelu.rst", "python/nn/_autosummary_functions/mlx.nn.gelu_approx.rst", "python/nn/_autosummary_functions/mlx.nn.gelu_fast_approx.rst", "python/nn/_autosummary_functions/mlx.nn.glu.rst", "python/nn/_autosummary_functions/mlx.nn.hard_shrink.rst", "python/nn/_autosummary_functions/mlx.nn.hard_tanh.rst", "python/nn/_autosummary_functions/mlx.nn.hardswish.rst", "python/nn/_autosummary_functions/mlx.nn.leaky_relu.rst", "python/nn/_autosummary_functions/mlx.nn.log_sigmoid.rst", "python/nn/_autosummary_functions/mlx.nn.log_softmax.rst", "python/nn/_autosummary_functions/mlx.nn.losses.binary_cross_entropy.rst", "python/nn/_autosummary_functions/mlx.nn.losses.cosine_similarity_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.cross_entropy.rst", "python/nn/_autosummary_functions/mlx.nn.losses.gaussian_nll_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.hinge_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.huber_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.kl_div_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.l1_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.log_cosh_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.margin_ranking_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.mse_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.nll_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.smooth_l1_loss.rst", "python/nn/_autosummary_functions/mlx.nn.losses.triplet_loss.rst", "python/nn/_autosummary_functions/mlx.nn.mish.rst", "python/nn/_autosummary_functions/mlx.nn.prelu.rst", "python/nn/_autosummary_functions/mlx.nn.relu.rst", "python/nn/_autosummary_functions/mlx.nn.relu6.rst", "python/nn/_autosummary_functions/mlx.nn.selu.rst", "python/nn/_autosummary_functions/mlx.nn.sigmoid.rst", "python/nn/_autosummary_functions/mlx.nn.silu.rst", "python/nn/_autosummary_functions/mlx.nn.softmax.rst", "python/nn/_autosummary_functions/mlx.nn.softmin.rst", "python/nn/_autosummary_functions/mlx.nn.softplus.rst", "python/nn/_autosummary_functions/mlx.nn.softshrink.rst", "python/nn/_autosummary_functions/mlx.nn.step.rst", "python/nn/_autosummary_functions/mlx.nn.tanh.rst", "python/nn/functions.rst", "python/nn/init.rst", "python/nn/layers.rst", "python/nn/losses.rst", "python/nn/module.rst", "python/ops.rst", "python/optimizers.rst", "python/optimizers/_autosummary/mlx.optimizers.AdaDelta.rst", "python/optimizers/_autosummary/mlx.optimizers.Adafactor.rst", "python/optimizers/_autosummary/mlx.optimizers.Adagrad.rst", "python/optimizers/_autosummary/mlx.optimizers.Adam.rst", "python/optimizers/_autosummary/mlx.optimizers.AdamW.rst", "python/optimizers/_autosummary/mlx.optimizers.Adamax.rst", "python/optimizers/_autosummary/mlx.optimizers.Lion.rst", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.apply_gradients.rst", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.init.rst", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.state.rst", "python/optimizers/_autosummary/mlx.optimizers.Optimizer.update.rst", "python/optimizers/_autosummary/mlx.optimizers.RMSprop.rst", "python/optimizers/_autosummary/mlx.optimizers.SGD.rst", "python/optimizers/_autosummary/mlx.optimizers.cosine_decay.rst", "python/optimizers/_autosummary/mlx.optimizers.exponential_decay.rst", "python/optimizers/_autosummary/mlx.optimizers.join_schedules.rst", "python/optimizers/_autosummary/mlx.optimizers.linear_schedule.rst", "python/optimizers/_autosummary/mlx.optimizers.step_decay.rst", "python/optimizers/common_optimizers.rst", "python/optimizers/optimizer.rst", "python/optimizers/schedulers.rst", "python/random.rst", "python/transforms.rst", "python/tree_utils.rst", "usage/compile.rst", "usage/distributed.rst", "usage/function_transforms.rst", "usage/indexing.rst", "usage/lazy_evaluation.rst", "usage/numpy.rst", "usage/quick_start.rst", "usage/saving_and_loading.rst", "usage/unified_memory.rst", "usage/using_streams.rst"], "indexentries": {"__init__() (array method)": [[30, "mlx.core.array.__init__", false]], "__init__() (custom_function method)": [[112, "mlx.core.custom_function.__init__", false]], "__init__() (device method)": [[9, "mlx.core.Device.__init__", false]], "__init__() (dtype method)": [[10, "mlx.core.Dtype.__init__", false]], "__init__() (dtypecategory method)": [[11, "mlx.core.DtypeCategory.__init__", false]], "__init__() (group method)": [[120, "mlx.core.distributed.Group.__init__", false]], "__init__() (stream method)": [[315, "mlx.core.Stream.__init__", false]], "abs (c++ function)": [[0, "_CPPv43absRK5array14StreamOrDevice", false]], "abs() (array method)": [[32, "mlx.core.array.abs", false]], "abs() (in module mlx.core)": [[12, "mlx.core.abs", false]], "adadelta (class in mlx.optimizers)": [[455, "mlx.optimizers.AdaDelta", false]], "adafactor (class in mlx.optimizers)": [[456, "mlx.optimizers.Adafactor", false]], "adagrad (class in mlx.optimizers)": [[457, "mlx.optimizers.Adagrad", false]], "adam (class in mlx.optimizers)": [[458, "mlx.optimizers.Adam", false]], "adamax (class in mlx.optimizers)": [[460, "mlx.optimizers.Adamax", false]], "adamw (class in mlx.optimizers)": [[459, "mlx.optimizers.AdamW", false]], "add (c++ function)": [[0, "_CPPv43addRK5arrayRK5array14StreamOrDevice", false]], "add() (in module mlx.core)": [[13, "mlx.core.add", false]], "addmm (c++ function)": [[0, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", false]], "addmm() (in module mlx.core)": [[14, "mlx.core.addmm", false]], "affine_quantize() (in module mlx.core.fast)": [[141, "mlx.core.fast.affine_quantize", false]], "alibi (class in mlx.nn)": [[325, "mlx.nn.ALiBi", false]], "all (c++ function)": [[0, "_CPPv43allRK5array14StreamOrDevice", false], [0, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv43allRK5arrayb14StreamOrDevice", false], [0, "_CPPv43allRK5arrayib14StreamOrDevice", false]], "all() (array method)": [[33, "mlx.core.array.all", false]], "all() (in module mlx.core)": [[15, "mlx.core.all", false]], "all_gather() (in module mlx.core.distributed)": [[121, "mlx.core.distributed.all_gather", false]], "all_sum() (in module mlx.core.distributed)": [[122, "mlx.core.distributed.all_sum", false]], "allclose (c++ function)": [[0, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", false]], "allclose() (in module mlx.core)": [[16, "mlx.core.allclose", false]], "any (c++ function)": [[0, "_CPPv43anyRK5array14StreamOrDevice", false], [0, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv43anyRK5arrayb14StreamOrDevice", false], [0, "_CPPv43anyRK5arrayib14StreamOrDevice", false]], "any() (array method)": [[34, "mlx.core.array.any", false]], "any() (in module mlx.core)": [[17, "mlx.core.any", false]], "apply() (module method)": [[358, "mlx.nn.Module.apply", false]], "apply_gradients() (optimizer method)": [[462, "mlx.optimizers.Optimizer.apply_gradients", false]], "apply_to_modules() (module method)": [[359, "mlx.nn.Module.apply_to_modules", false]], "arange (c++ function)": [[0, "_CPPv46aranged14StreamOrDevice", false], [0, "_CPPv46aranged5Dtype14StreamOrDevice", false], [0, "_CPPv46arangedd14StreamOrDevice", false], [0, "_CPPv46arangedd5Dtype14StreamOrDevice", false], [0, "_CPPv46arangeddd14StreamOrDevice", false], [0, "_CPPv46arangeddd5Dtype14StreamOrDevice", false], [0, "_CPPv46arangei14StreamOrDevice", false], [0, "_CPPv46arangeii14StreamOrDevice", false], [0, "_CPPv46arangeiii14StreamOrDevice", false]], "arange() (in module mlx.core)": [[18, "mlx.core.arange", false]], "arccos (c++ function)": [[0, "_CPPv46arccosRK5array14StreamOrDevice", false]], "arccos() (in module mlx.core)": [[19, "mlx.core.arccos", false]], "arccosh (c++ function)": [[0, "_CPPv47arccoshRK5array14StreamOrDevice", false]], "arccosh() (in module mlx.core)": [[20, "mlx.core.arccosh", false]], "arcsin (c++ function)": [[0, "_CPPv46arcsinRK5array14StreamOrDevice", false]], "arcsin() (in module mlx.core)": [[21, "mlx.core.arcsin", false]], "arcsinh (c++ function)": [[0, "_CPPv47arcsinhRK5array14StreamOrDevice", false]], "arcsinh() (in module mlx.core)": [[22, "mlx.core.arcsinh", false]], "arctan (c++ function)": [[0, "_CPPv46arctanRK5array14StreamOrDevice", false]], "arctan() (in module mlx.core)": [[23, "mlx.core.arctan", false]], "arctan2 (c++ function)": [[0, "_CPPv47arctan2RK5arrayRK5array14StreamOrDevice", false]], "arctan2() (in module mlx.core)": [[24, "mlx.core.arctan2", false]], "arctanh (c++ function)": [[0, "_CPPv47arctanhRK5array14StreamOrDevice", false]], "arctanh() (in module mlx.core)": [[25, "mlx.core.arctanh", false]], "argmax (c++ function)": [[0, "_CPPv46argmaxRK5array14StreamOrDevice", false], [0, "_CPPv46argmaxRK5arrayb14StreamOrDevice", false], [0, "_CPPv46argmaxRK5arrayib14StreamOrDevice", false]], "argmax() (array method)": [[35, "mlx.core.array.argmax", false]], "argmax() (in module mlx.core)": [[26, "mlx.core.argmax", false]], "argmin (c++ function)": [[0, "_CPPv46argminRK5array14StreamOrDevice", false], [0, "_CPPv46argminRK5arrayb14StreamOrDevice", false], [0, "_CPPv46argminRK5arrayib14StreamOrDevice", false]], "argmin() (array method)": [[36, "mlx.core.array.argmin", false]], "argmin() (in module mlx.core)": [[27, "mlx.core.argmin", false]], "argpartition (c++ function)": [[0, "_CPPv412argpartitionRK5arrayi14StreamOrDevice", false], [0, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", false]], "argpartition() (in module mlx.core)": [[28, "mlx.core.argpartition", false]], "argsort (c++ function)": [[0, "_CPPv47argsortRK5array14StreamOrDevice", false], [0, "_CPPv47argsortRK5arrayi14StreamOrDevice", false]], "argsort() (in module mlx.core)": [[29, "mlx.core.argsort", false]], "array (class in mlx.core)": [[30, "mlx.core.array", false]], "array_equal (c++ function)": [[0, "_CPPv411array_equalRK5arrayRK5array14StreamOrDevice", false], [0, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", false]], "array_equal() (in module mlx.core)": [[82, "mlx.core.array_equal", false]], "as_strided (c++ function)": [[0, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", false]], "as_strided() (in module mlx.core)": [[83, "mlx.core.as_strided", false]], "astype (c++ function)": [[0, "_CPPv46astype5array5Dtype14StreamOrDevice", false]], "astype() (array method)": [[37, "mlx.core.array.astype", false]], "at (array property)": [[38, "mlx.core.array.at", false]], "atleast_1d (c++ function)": [[0, "_CPPv410atleast_1dRK5array14StreamOrDevice", false], [0, "_CPPv410atleast_1dRKNSt6vectorI5arrayEE14StreamOrDevice", false]], "atleast_1d() (in module mlx.core)": [[84, "mlx.core.atleast_1d", false]], "atleast_2d (c++ function)": [[0, "_CPPv410atleast_2dRK5array14StreamOrDevice", false], [0, "_CPPv410atleast_2dRKNSt6vectorI5arrayEE14StreamOrDevice", false]], "atleast_2d() (in module mlx.core)": [[85, "mlx.core.atleast_2d", false]], "atleast_3d (c++ function)": [[0, "_CPPv410atleast_3dRK5array14StreamOrDevice", false], [0, "_CPPv410atleast_3dRKNSt6vectorI5arrayEE14StreamOrDevice", false]], "atleast_3d() (in module mlx.core)": [[86, "mlx.core.atleast_3d", false]], "avgpool1d (class in mlx.nn)": [[326, "mlx.nn.AvgPool1d", false]], "avgpool2d (class in mlx.nn)": [[327, "mlx.nn.AvgPool2d", false]], "batchnorm (class in mlx.nn)": [[328, "mlx.nn.BatchNorm", false]], "bernoulli() (in module mlx.core.random)": [[240, "mlx.core.random.bernoulli", false]], "binary_cross_entropy (class in mlx.nn.losses)": [[421, "mlx.nn.losses.binary_cross_entropy", false]], "bitwise_and (c++ function)": [[0, "_CPPv411bitwise_andRK5arrayRK5array14StreamOrDevice", false]], "bitwise_and() (in module mlx.core)": [[87, "mlx.core.bitwise_and", false]], "bitwise_or (c++ function)": [[0, "_CPPv410bitwise_orRK5arrayRK5array14StreamOrDevice", false]], "bitwise_or() (in module mlx.core)": [[88, "mlx.core.bitwise_or", false]], "bitwise_xor (c++ function)": [[0, "_CPPv411bitwise_xorRK5arrayRK5array14StreamOrDevice", false]], "bitwise_xor() (in module mlx.core)": [[89, "mlx.core.bitwise_xor", false]], "block_masked_mm (c++ function)": [[0, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", false]], "block_masked_mm() (in module mlx.core)": [[90, "mlx.core.block_masked_mm", false]], "broadcast_arrays (c++ function)": [[0, "_CPPv416broadcast_arraysRKNSt6vectorI5arrayEE14StreamOrDevice", false]], "broadcast_to (c++ function)": [[0, "_CPPv412broadcast_toRK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "broadcast_to() (in module mlx.core)": [[91, "mlx.core.broadcast_to", false]], "categorical() (in module mlx.core.random)": [[241, "mlx.core.random.categorical", false]], "ceil (c++ function)": [[0, "_CPPv44ceilRK5array14StreamOrDevice", false]], "ceil() (in module mlx.core)": [[92, "mlx.core.ceil", false]], "celu (class in mlx.nn)": [[329, "mlx.nn.CELU", false], [409, "mlx.nn.celu", false]], "children() (module method)": [[360, "mlx.nn.Module.children", false]], "cholesky() (in module mlx.core.linalg)": [[183, "mlx.core.linalg.cholesky", false]], "cholesky_inv() (in module mlx.core.linalg)": [[184, "mlx.core.linalg.cholesky_inv", false]], "clear_cache() (in module mlx.core.metal)": [[209, "mlx.core.metal.clear_cache", false]], "clip (c++ function)": [[0, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", false]], "clip() (in module mlx.core)": [[93, "mlx.core.clip", false]], "clip_grad_norm() (in module mlx.optimizers)": [[309, "mlx.optimizers.clip_grad_norm", false]], "compile() (in module mlx.core)": [[94, "mlx.core.compile", false]], "concatenate (c++ function)": [[0, "_CPPv411concatenateRKNSt6vectorI5arrayEE14StreamOrDevice", false], [0, "_CPPv411concatenateRKNSt6vectorI5arrayEEi14StreamOrDevice", false]], "concatenate() (in module mlx.core)": [[95, "mlx.core.concatenate", false]], "conj() (array method)": [[39, "mlx.core.array.conj", false]], "conj() (in module mlx.core)": [[96, "mlx.core.conj", false]], "conjugate (c++ function)": [[0, "_CPPv49conjugateRK5array14StreamOrDevice", false]], "conjugate() (in module mlx.core)": [[97, "mlx.core.conjugate", false]], "constant() (in module mlx.nn.init)": [[401, "mlx.nn.init.constant", false]], "conv1d (c++ function)": [[0, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", false]], "conv1d (class in mlx.nn)": [[330, "mlx.nn.Conv1d", false]], "conv1d() (in module mlx.core)": [[98, "mlx.core.conv1d", false]], "conv2d (c++ function)": [[0, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", false]], "conv2d (class in mlx.nn)": [[331, "mlx.nn.Conv2d", false]], "conv2d() (in module mlx.core)": [[99, "mlx.core.conv2d", false]], "conv3d (c++ function)": [[0, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", false]], "conv3d (class in mlx.nn)": [[332, "mlx.nn.Conv3d", false]], "conv3d() (in module mlx.core)": [[100, "mlx.core.conv3d", false]], "conv_general (c++ function)": [[0, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", false], [0, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", false]], "conv_general() (in module mlx.core)": [[101, "mlx.core.conv_general", false]], "conv_transpose1d (c++ function)": [[0, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", false]], "conv_transpose1d() (in module mlx.core)": [[102, "mlx.core.conv_transpose1d", false]], "conv_transpose2d (c++ function)": [[0, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", false]], "conv_transpose2d() (in module mlx.core)": [[103, "mlx.core.conv_transpose2d", false]], "conv_transpose3d (c++ function)": [[0, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", false]], "conv_transpose3d() (in module mlx.core)": [[104, "mlx.core.conv_transpose3d", false]], "convolve() (in module mlx.core)": [[105, "mlx.core.convolve", false]], "convtranspose1d (class in mlx.nn)": [[333, "mlx.nn.ConvTranspose1d", false]], "convtranspose2d (class in mlx.nn)": [[334, "mlx.nn.ConvTranspose2d", false]], "convtranspose3d (class in mlx.nn)": [[335, "mlx.nn.ConvTranspose3d", false]], "copy (c++ function)": [[0, "_CPPv44copy5array14StreamOrDevice", false]], "cos (c++ function)": [[0, "_CPPv43cosRK5array14StreamOrDevice", false]], "cos() (array method)": [[40, "mlx.core.array.cos", false]], "cos() (in module mlx.core)": [[106, "mlx.core.cos", false]], "cosh (c++ function)": [[0, "_CPPv44coshRK5array14StreamOrDevice", false]], "cosh() (in module mlx.core)": [[107, "mlx.core.cosh", false]], "cosine_decay() (in module mlx.optimizers)": [[468, "mlx.optimizers.cosine_decay", false]], "cosine_similarity_loss (class in mlx.nn.losses)": [[422, "mlx.nn.losses.cosine_similarity_loss", false]], "cross() (in module mlx.core.linalg)": [[185, "mlx.core.linalg.cross", false]], "cross_entropy (class in mlx.nn.losses)": [[423, "mlx.nn.losses.cross_entropy", false]], "cummax (c++ function)": [[0, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", false]], "cummax() (array method)": [[41, "mlx.core.array.cummax", false]], "cummax() (in module mlx.core)": [[108, "mlx.core.cummax", false]], "cummin (c++ function)": [[0, "_CPPv46cumminRK5arrayibb14StreamOrDevice", false]], "cummin() (array method)": [[42, "mlx.core.array.cummin", false]], "cummin() (in module mlx.core)": [[109, "mlx.core.cummin", false]], "cumprod (c++ function)": [[0, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", false]], "cumprod() (array method)": [[43, "mlx.core.array.cumprod", false]], "cumprod() (in module mlx.core)": [[110, "mlx.core.cumprod", false]], "cumsum (c++ function)": [[0, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", false]], "cumsum() (array method)": [[44, "mlx.core.array.cumsum", false]], "cumsum() (in module mlx.core)": [[111, "mlx.core.cumsum", false]], "custom_function (class in mlx.core)": [[112, "mlx.core.custom_function", false]], "default_device() (in module mlx.core)": [[113, "mlx.core.default_device", false]], "default_stream() (in module mlx.core)": [[114, "mlx.core.default_stream", false]], "degrees (c++ function)": [[0, "_CPPv47degreesRK5array14StreamOrDevice", false]], "degrees() (in module mlx.core)": [[115, "mlx.core.degrees", false]], "depends (c++ function)": [[0, "_CPPv47dependsRKNSt6vectorI5arrayEERKNSt6vectorI5arrayEE", false]], "dequantize (c++ function)": [[0, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", false]], "dequantize() (in module mlx.core)": [[116, "mlx.core.dequantize", false]], "device (class in mlx.core)": [[9, "mlx.core.Device", false]], "device_info() (in module mlx.core.metal)": [[210, "mlx.core.metal.device_info", false]], "diag (c++ function)": [[0, "_CPPv44diagRK5arrayi14StreamOrDevice", false]], "diag() (array method)": [[45, "mlx.core.array.diag", false]], "diag() (in module mlx.core)": [[117, "mlx.core.diag", false]], "diagonal (c++ function)": [[0, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", false]], "diagonal() (array method)": [[46, "mlx.core.array.diagonal", false]], "diagonal() (in module mlx.core)": [[118, "mlx.core.diagonal", false]], "disable_compile() (in module mlx.core)": [[119, "mlx.core.disable_compile", false]], "divide (c++ function)": [[0, "_CPPv46divideRK5arrayRK5array14StreamOrDevice", false]], "divide() (in module mlx.core)": [[128, "mlx.core.divide", false]], "divmod (c++ function)": [[0, "_CPPv46divmodRK5arrayRK5array14StreamOrDevice", false]], "divmod() (in module mlx.core)": [[129, "mlx.core.divmod", false]], "dropout (class in mlx.nn)": [[336, "mlx.nn.Dropout", false]], "dropout2d (class in mlx.nn)": [[337, "mlx.nn.Dropout2d", false]], "dropout3d (class in mlx.nn)": [[338, "mlx.nn.Dropout3d", false]], "dtype (array property)": [[47, "mlx.core.array.dtype", false]], "dtype (class in mlx.core)": [[10, "mlx.core.Dtype", false]], "dtypecategory (class in mlx.core)": [[11, "mlx.core.DtypeCategory", false]], "eigh() (in module mlx.core.linalg)": [[186, "mlx.core.linalg.eigh", false]], "eigvalsh() (in module mlx.core.linalg)": [[187, "mlx.core.linalg.eigvalsh", false]], "einsum() (in module mlx.core)": [[130, "mlx.core.einsum", false]], "einsum_path() (in module mlx.core)": [[131, "mlx.core.einsum_path", false]], "elu (class in mlx.nn)": [[339, "mlx.nn.ELU", false], [410, "mlx.nn.elu", false]], "embedding (class in mlx.nn)": [[340, "mlx.nn.Embedding", false]], "enable_compile() (in module mlx.core)": [[132, "mlx.core.enable_compile", false]], "equal (c++ function)": [[0, "_CPPv45equalRK5arrayRK5array14StreamOrDevice", false]], "equal() (in module mlx.core)": [[133, "mlx.core.equal", false]], "erf (c++ function)": [[0, "_CPPv43erfRK5array14StreamOrDevice", false]], "erf() (in module mlx.core)": [[134, "mlx.core.erf", false]], "erfinv (c++ function)": [[0, "_CPPv46erfinvRK5array14StreamOrDevice", false]], "erfinv() (in module mlx.core)": [[135, "mlx.core.erfinv", false]], "eval() (in module mlx.core)": [[136, "mlx.core.eval", false]], "eval() (module method)": [[361, "mlx.nn.Module.eval", false]], "exp (c++ function)": [[0, "_CPPv43expRK5array14StreamOrDevice", false]], "exp() (array method)": [[48, "mlx.core.array.exp", false]], "exp() (in module mlx.core)": [[137, "mlx.core.exp", false]], "expand_dims (c++ function)": [[0, "_CPPv411expand_dimsRK5arrayRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv411expand_dimsRK5arrayi14StreamOrDevice", false]], "expand_dims() (in module mlx.core)": [[138, "mlx.core.expand_dims", false]], "expm1 (c++ function)": [[0, "_CPPv45expm1RK5array14StreamOrDevice", false]], "expm1() (in module mlx.core)": [[139, "mlx.core.expm1", false]], "exponential_decay() (in module mlx.optimizers)": [[469, "mlx.optimizers.exponential_decay", false]], "eye (c++ function)": [[0, "_CPPv43eyei14StreamOrDevice", false], [0, "_CPPv43eyei5Dtype14StreamOrDevice", false], [0, "_CPPv43eyeii14StreamOrDevice", false], [0, "_CPPv43eyeiii14StreamOrDevice", false], [0, "_CPPv43eyeiii5Dtype14StreamOrDevice", false]], "eye() (in module mlx.core)": [[140, "mlx.core.eye", false]], "fft() (in module mlx.core.fft)": [[147, "mlx.core.fft.fft", false]], "fft2() (in module mlx.core.fft)": [[148, "mlx.core.fft.fft2", false]], "fftn() (in module mlx.core.fft)": [[149, "mlx.core.fft.fftn", false]], "filter_and_map() (module method)": [[362, "mlx.nn.Module.filter_and_map", false]], "flatten (c++ function)": [[0, "_CPPv47flattenRK5array14StreamOrDevice", false], [0, "_CPPv47flattenRK5arrayii14StreamOrDevice", false]], "flatten() (array method)": [[49, "mlx.core.array.flatten", false]], "flatten() (in module mlx.core)": [[159, "mlx.core.flatten", false]], "floor (c++ function)": [[0, "_CPPv45floorRK5array14StreamOrDevice", false]], "floor() (in module mlx.core)": [[160, "mlx.core.floor", false]], "floor_divide (c++ function)": [[0, "_CPPv412floor_divideRK5arrayRK5array14StreamOrDevice", false]], "floor_divide() (in module mlx.core)": [[161, "mlx.core.floor_divide", false]], "freeze() (module method)": [[363, "mlx.nn.Module.freeze", false]], "full (c++ function)": [[0, "_CPPv44fullNSt6vectorIiEE5array14StreamOrDevice", false], [0, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", false], [0, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", false], [0, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", false]], "full() (in module mlx.core)": [[162, "mlx.core.full", false]], "gather (c++ function)": [[0, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", false]], "gather_mm (c++ function)": [[0, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", false]], "gather_mm() (in module mlx.core)": [[163, "mlx.core.gather_mm", false]], "gather_qmm (c++ function)": [[0, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", false]], "gather_qmm() (in module mlx.core)": [[164, "mlx.core.gather_qmm", false]], "gaussian_nll_loss (class in mlx.nn.losses)": [[424, "mlx.nn.losses.gaussian_nll_loss", false]], "gelu (class in mlx.nn)": [[341, "mlx.nn.GELU", false], [411, "mlx.nn.gelu", false]], "gelu_approx (class in mlx.nn)": [[412, "mlx.nn.gelu_approx", false]], "gelu_fast_approx (class in mlx.nn)": [[413, "mlx.nn.gelu_fast_approx", false]], "get_active_memory() (in module mlx.core.metal)": [[211, "mlx.core.metal.get_active_memory", false]], "get_cache_memory() (in module mlx.core.metal)": [[212, "mlx.core.metal.get_cache_memory", false]], "get_peak_memory() (in module mlx.core.metal)": [[213, "mlx.core.metal.get_peak_memory", false]], "glorot_normal() (in module mlx.nn.init)": [[402, "mlx.nn.init.glorot_normal", false]], "glorot_uniform() (in module mlx.nn.init)": [[403, "mlx.nn.init.glorot_uniform", false]], "glu (class in mlx.nn)": [[342, "mlx.nn.GLU", false], [414, "mlx.nn.glu", false]], "grad() (in module mlx.core)": [[165, "mlx.core.grad", false]], "greater (c++ function)": [[0, "_CPPv47greaterRK5arrayRK5array14StreamOrDevice", false]], "greater() (in module mlx.core)": [[166, "mlx.core.greater", false]], "greater_equal (c++ function)": [[0, "_CPPv413greater_equalRK5arrayRK5array14StreamOrDevice", false]], "greater_equal() (in module mlx.core)": [[167, "mlx.core.greater_equal", false]], "group (class in mlx.core.distributed)": [[120, "mlx.core.distributed.Group", false]], "groupnorm (class in mlx.nn)": [[344, "mlx.nn.GroupNorm", false]], "gru (class in mlx.nn)": [[343, "mlx.nn.GRU", false]], "gumbel() (in module mlx.core.random)": [[242, "mlx.core.random.gumbel", false]], "hadamard_transform (c++ function)": [[0, "_CPPv418hadamard_transformRK5arrayNSt8optionalIfEE14StreamOrDevice", false]], "hadamard_transform() (in module mlx.core)": [[168, "mlx.core.hadamard_transform", false]], "hard_shrink (class in mlx.nn)": [[415, "mlx.nn.hard_shrink", false]], "hard_tanh (class in mlx.nn)": [[416, "mlx.nn.hard_tanh", false]], "hardshrink (class in mlx.nn)": [[345, "mlx.nn.HardShrink", false]], "hardswish (class in mlx.nn)": [[347, "mlx.nn.Hardswish", false], [417, "mlx.nn.hardswish", false]], "hardtanh (class in mlx.nn)": [[346, "mlx.nn.HardTanh", false]], "he_normal() (in module mlx.nn.init)": [[404, "mlx.nn.init.he_normal", false]], "he_uniform() (in module mlx.nn.init)": [[405, "mlx.nn.init.he_uniform", false]], "hinge_loss (class in mlx.nn.losses)": [[425, "mlx.nn.losses.hinge_loss", false]], "huber_loss (class in mlx.nn.losses)": [[426, "mlx.nn.losses.huber_loss", false]], "identity (c++ function)": [[0, "_CPPv48identityi14StreamOrDevice", false], [0, "_CPPv48identityi5Dtype14StreamOrDevice", false]], "identity() (in module mlx.core)": [[169, "mlx.core.identity", false]], "identity() (in module mlx.nn.init)": [[406, "mlx.nn.init.identity", false]], "ifft() (in module mlx.core.fft)": [[150, "mlx.core.fft.ifft", false]], "ifft2() (in module mlx.core.fft)": [[151, "mlx.core.fft.ifft2", false]], "ifftn() (in module mlx.core.fft)": [[152, "mlx.core.fft.ifftn", false]], "imag (c++ function)": [[0, "_CPPv44imagRK5array14StreamOrDevice", false]], "imag() (in module mlx.core)": [[170, "mlx.core.imag", false]], "init() (in module mlx.core.distributed)": [[123, "mlx.core.distributed.init", false]], "init() (optimizer method)": [[463, "mlx.optimizers.Optimizer.init", false]], "inner (c++ function)": [[0, "_CPPv45innerRK5arrayRK5array14StreamOrDevice", false]], "inner() (in module mlx.core)": [[171, "mlx.core.inner", false]], "instancenorm (class in mlx.nn)": [[348, "mlx.nn.InstanceNorm", false]], "inv() (in module mlx.core.linalg)": [[188, "mlx.core.linalg.inv", false]], "irfft() (in module mlx.core.fft)": [[153, "mlx.core.fft.irfft", false]], "irfft2() (in module mlx.core.fft)": [[154, "mlx.core.fft.irfft2", false]], "irfftn() (in module mlx.core.fft)": [[155, "mlx.core.fft.irfftn", false]], "is_available() (in module mlx.core.distributed)": [[124, "mlx.core.distributed.is_available", false]], "is_available() (in module mlx.core.metal)": [[214, "mlx.core.metal.is_available", false]], "isclose (c++ function)": [[0, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", false]], "isclose() (in module mlx.core)": [[172, "mlx.core.isclose", false]], "isfinite (c++ function)": [[0, "_CPPv48isfiniteRK5array14StreamOrDevice", false]], "isfinite() (in module mlx.core)": [[173, "mlx.core.isfinite", false]], "isinf (c++ function)": [[0, "_CPPv45isinfRK5array14StreamOrDevice", false]], "isinf() (in module mlx.core)": [[174, "mlx.core.isinf", false]], "isnan (c++ function)": [[0, "_CPPv45isnanRK5array14StreamOrDevice", false]], "isnan() (in module mlx.core)": [[175, "mlx.core.isnan", false]], "isneginf (c++ function)": [[0, "_CPPv48isneginfRK5array14StreamOrDevice", false]], "isneginf() (in module mlx.core)": [[176, "mlx.core.isneginf", false]], "isposinf (c++ function)": [[0, "_CPPv48isposinfRK5array14StreamOrDevice", false]], "isposinf() (in module mlx.core)": [[177, "mlx.core.isposinf", false]], "issubdtype() (in module mlx.core)": [[178, "mlx.core.issubdtype", false]], "item() (array method)": [[50, "mlx.core.array.item", false]], "itemsize (array property)": [[51, "mlx.core.array.itemsize", false]], "join_schedules() (in module mlx.optimizers)": [[470, "mlx.optimizers.join_schedules", false]], "jvp() (in module mlx.core)": [[179, "mlx.core.jvp", false]], "key() (in module mlx.core.random)": [[243, "mlx.core.random.key", false]], "kl_div_loss (class in mlx.nn.losses)": [[427, "mlx.nn.losses.kl_div_loss", false]], "l1_loss (class in mlx.nn.losses)": [[428, "mlx.nn.losses.l1_loss", false]], "laplace() (in module mlx.core.random)": [[244, "mlx.core.random.laplace", false]], "layer_norm() (in module mlx.core.fast)": [[142, "mlx.core.fast.layer_norm", false]], "layernorm (class in mlx.nn)": [[350, "mlx.nn.LayerNorm", false]], "leaf_modules() (module method)": [[364, "mlx.nn.Module.leaf_modules", false]], "leaky_relu (class in mlx.nn)": [[418, "mlx.nn.leaky_relu", false]], "leakyrelu (class in mlx.nn)": [[351, "mlx.nn.LeakyReLU", false]], "left_shift (c++ function)": [[0, "_CPPv410left_shiftRK5arrayRK5array14StreamOrDevice", false]], "left_shift() (in module mlx.core)": [[180, "mlx.core.left_shift", false]], "less (c++ function)": [[0, "_CPPv44lessRK5arrayRK5array14StreamOrDevice", false]], "less() (in module mlx.core)": [[181, "mlx.core.less", false]], "less_equal (c++ function)": [[0, "_CPPv410less_equalRK5arrayRK5array14StreamOrDevice", false]], "less_equal() (in module mlx.core)": [[182, "mlx.core.less_equal", false]], "linear (class in mlx.nn)": [[352, "mlx.nn.Linear", false]], "linear_schedule() (in module mlx.optimizers)": [[471, "mlx.optimizers.linear_schedule", false]], "linspace (c++ function)": [[0, "_CPPv48linspaceddi5Dtype14StreamOrDevice", false]], "linspace() (in module mlx.core)": [[193, "mlx.core.linspace", false]], "lion (class in mlx.optimizers)": [[461, "mlx.optimizers.Lion", false]], "load() (in module mlx.core)": [[194, "mlx.core.load", false]], "load_weights() (module method)": [[365, "mlx.nn.Module.load_weights", false]], "log (c++ function)": [[0, "_CPPv43logRK5array14StreamOrDevice", false]], "log() (array method)": [[52, "mlx.core.array.log", false]], "log() (in module mlx.core)": [[195, "mlx.core.log", false]], "log10 (c++ function)": [[0, "_CPPv45log10RK5array14StreamOrDevice", false]], "log10() (array method)": [[53, "mlx.core.array.log10", false]], "log10() (in module mlx.core)": [[196, "mlx.core.log10", false]], "log1p (c++ function)": [[0, "_CPPv45log1pRK5array14StreamOrDevice", false]], "log1p() (array method)": [[54, "mlx.core.array.log1p", false]], "log1p() (in module mlx.core)": [[197, "mlx.core.log1p", false]], "log2 (c++ function)": [[0, "_CPPv44log2RK5array14StreamOrDevice", false]], "log2() (array method)": [[55, "mlx.core.array.log2", false]], "log2() (in module mlx.core)": [[198, "mlx.core.log2", false]], "log_cosh_loss (class in mlx.nn.losses)": [[429, "mlx.nn.losses.log_cosh_loss", false]], "log_sigmoid (class in mlx.nn)": [[419, "mlx.nn.log_sigmoid", false]], "log_softmax (class in mlx.nn)": [[420, "mlx.nn.log_softmax", false]], "logaddexp (c++ function)": [[0, "_CPPv49logaddexpRK5arrayRK5array14StreamOrDevice", false]], "logaddexp() (in module mlx.core)": [[199, "mlx.core.logaddexp", false]], "logical_and (c++ function)": [[0, "_CPPv411logical_andRK5arrayRK5array14StreamOrDevice", false]], "logical_and() (in module mlx.core)": [[200, "mlx.core.logical_and", false]], "logical_not (c++ function)": [[0, "_CPPv411logical_notRK5array14StreamOrDevice", false]], "logical_not() (in module mlx.core)": [[201, "mlx.core.logical_not", false]], "logical_or (c++ function)": [[0, "_CPPv410logical_orRK5arrayRK5array14StreamOrDevice", false]], "logical_or() (in module mlx.core)": [[202, "mlx.core.logical_or", false]], "logsigmoid (class in mlx.nn)": [[353, "mlx.nn.LogSigmoid", false]], "logsoftmax (class in mlx.nn)": [[354, "mlx.nn.LogSoftmax", false]], "logsumexp (c++ function)": [[0, "_CPPv49logsumexpRK5array14StreamOrDevice", false], [0, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv49logsumexpRK5arrayb14StreamOrDevice", false], [0, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", false]], "logsumexp() (array method)": [[56, "mlx.core.array.logsumexp", false]], "logsumexp() (in module mlx.core)": [[203, "mlx.core.logsumexp", false]], "lstm (class in mlx.nn)": [[349, "mlx.nn.LSTM", false]], "margin_ranking_loss (class in mlx.nn.losses)": [[430, "mlx.nn.losses.margin_ranking_loss", false]], "matmul (c++ function)": [[0, "_CPPv46matmulRK5arrayRK5array14StreamOrDevice", false]], "matmul() (in module mlx.core)": [[204, "mlx.core.matmul", false]], "max (c++ function)": [[0, "_CPPv43maxRK5array14StreamOrDevice", false], [0, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv43maxRK5arrayb14StreamOrDevice", false], [0, "_CPPv43maxRK5arrayib14StreamOrDevice", false]], "max() (array method)": [[57, "mlx.core.array.max", false]], "max() (in module mlx.core)": [[205, "mlx.core.max", false]], "maximum (c++ function)": [[0, "_CPPv47maximumRK5arrayRK5array14StreamOrDevice", false]], "maximum() (in module mlx.core)": [[206, "mlx.core.maximum", false]], "maxpool1d (class in mlx.nn)": [[355, "mlx.nn.MaxPool1d", false]], "maxpool2d (class in mlx.nn)": [[356, "mlx.nn.MaxPool2d", false]], "mean (c++ function)": [[0, "_CPPv44meanRK5array14StreamOrDevice", false], [0, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv44meanRK5arrayb14StreamOrDevice", false], [0, "_CPPv44meanRK5arrayib14StreamOrDevice", false]], "mean() (array method)": [[58, "mlx.core.array.mean", false]], "mean() (in module mlx.core)": [[207, "mlx.core.mean", false]], "meshgrid (c++ function)": [[0, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", false]], "meshgrid() (in module mlx.core)": [[208, "mlx.core.meshgrid", false]], "metal_kernel() (in module mlx.core.fast)": [[143, "mlx.core.fast.metal_kernel", false]], "min (c++ function)": [[0, "_CPPv43minRK5array14StreamOrDevice", false], [0, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv43minRK5arrayb14StreamOrDevice", false], [0, "_CPPv43minRK5arrayib14StreamOrDevice", false]], "min() (array method)": [[59, "mlx.core.array.min", false]], "min() (in module mlx.core)": [[221, "mlx.core.min", false]], "minimum (c++ function)": [[0, "_CPPv47minimumRK5arrayRK5array14StreamOrDevice", false]], "minimum() (in module mlx.core)": [[222, "mlx.core.minimum", false]], "mish (class in mlx.nn)": [[357, "mlx.nn.Mish", false], [435, "mlx.nn.mish", false]], "module (class in mlx.nn)": [[452, "mlx.nn.Module", false]], "modules() (module method)": [[366, "mlx.nn.Module.modules", false]], "moveaxis (c++ function)": [[0, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", false]], "moveaxis() (array method)": [[60, "mlx.core.array.moveaxis", false]], "moveaxis() (in module mlx.core)": [[223, "mlx.core.moveaxis", false]], "mse_loss (class in mlx.nn.losses)": [[431, "mlx.nn.losses.mse_loss", false]], "multiheadattention (class in mlx.nn)": [[378, "mlx.nn.MultiHeadAttention", false]], "multiply (c++ function)": [[0, "_CPPv48multiplyRK5arrayRK5array14StreamOrDevice", false]], "multiply() (in module mlx.core)": [[224, "mlx.core.multiply", false]], "multivariate_normal() (in module mlx.core.random)": [[245, "mlx.core.random.multivariate_normal", false]], "named_modules() (module method)": [[367, "mlx.nn.Module.named_modules", false]], "nan_to_num (c++ function)": [[0, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", false]], "nan_to_num() (in module mlx.core)": [[225, "mlx.core.nan_to_num", false]], "nbytes (array property)": [[61, "mlx.core.array.nbytes", false]], "ndim (array property)": [[62, "mlx.core.array.ndim", false]], "negative (c++ function)": [[0, "_CPPv48negativeRK5array14StreamOrDevice", false]], "negative() (in module mlx.core)": [[226, "mlx.core.negative", false]], "new_stream() (in module mlx.core)": [[227, "mlx.core.new_stream", false]], "nll_loss (class in mlx.nn.losses)": [[432, "mlx.nn.losses.nll_loss", false]], "norm() (in module mlx.core.linalg)": [[189, "mlx.core.linalg.norm", false]], "normal() (in module mlx.core.random)": [[246, "mlx.core.random.normal", false]], "normal() (in module mlx.nn.init)": [[407, "mlx.nn.init.normal", false]], "not_equal (c++ function)": [[0, "_CPPv49not_equalRK5arrayRK5array14StreamOrDevice", false]], "not_equal() (in module mlx.core)": [[228, "mlx.core.not_equal", false]], "number_of_elements (c++ function)": [[0, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", false]], "ones (c++ function)": [[0, "_CPPv44onesRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv44onesRKNSt6vectorIiEE5Dtype14StreamOrDevice", false]], "ones() (in module mlx.core)": [[229, "mlx.core.ones", false]], "ones_like (c++ function)": [[0, "_CPPv49ones_likeRK5array14StreamOrDevice", false]], "ones_like() (in module mlx.core)": [[230, "mlx.core.ones_like", false]], "operator!= (c++ function)": [[0, "_CPPv4I0Ene5array1TRK5array", false], [0, "_CPPv4I0Ene5arrayRK5array1T", false], [0, "_CPPv4neRK5arrayRK5array", false]], "operator% (c++ function)": [[0, "_CPPv4I0Erm5array1TRK5array", false], [0, "_CPPv4I0Erm5arrayRK5array1T", false], [0, "_CPPv4rmRK5arrayRK5array", false]], "operator& (c++ function)": [[0, "_CPPv4anRK5arrayRK5array", false]], "operator&& (c++ function)": [[0, "_CPPv4aaRK5arrayRK5array", false]], "operator* (c++ function)": [[0, "_CPPv4I0Eml5array1TRK5array", false], [0, "_CPPv4I0Eml5arrayRK5array1T", false], [0, "_CPPv4mlRK5arrayRK5array", false]], "operator+ (c++ function)": [[0, "_CPPv4I0Epl5array1TRK5array", false], [0, "_CPPv4I0Epl5arrayRK5array1T", false], [0, "_CPPv4plRK5arrayRK5array", false]], "operator- (c++ function)": [[0, "_CPPv4I0Emi5array1TRK5array", false], [0, "_CPPv4I0Emi5arrayRK5array1T", false], [0, "_CPPv4miRK5array", false], [0, "_CPPv4miRK5arrayRK5array", false]], "operator/ (c++ function)": [[0, "_CPPv4dvRK5arrayRK5array", false], [0, "_CPPv4dvRK5arrayd", false], [0, "_CPPv4dvdRK5array", false]], "operator< (c++ function)": [[0, "_CPPv4I0Elt5array1TRK5array", false], [0, "_CPPv4I0Elt5arrayRK5array1T", false], [0, "_CPPv4ltRK5arrayRK5array", false]], "operator<< (c++ function)": [[0, "_CPPv4lsRK5arrayRK5array", false]], "operator<= (c++ function)": [[0, "_CPPv4I0Ele5array1TRK5array", false], [0, "_CPPv4I0Ele5arrayRK5array1T", false], [0, "_CPPv4leRK5arrayRK5array", false]], "operator== (c++ function)": [[0, "_CPPv4I0Eeq5array1TRK5array", false], [0, "_CPPv4I0Eeq5arrayRK5array1T", false], [0, "_CPPv4eqRK5arrayRK5array", false]], "operator> (c++ function)": [[0, "_CPPv4I0Egt5array1TRK5array", false], [0, "_CPPv4I0Egt5arrayRK5array1T", false], [0, "_CPPv4gtRK5arrayRK5array", false]], "operator>= (c++ function)": [[0, "_CPPv4I0Ege5array1TRK5array", false], [0, "_CPPv4I0Ege5arrayRK5array1T", false], [0, "_CPPv4geRK5arrayRK5array", false]], "operator>> (c++ function)": [[0, "_CPPv4rsRK5arrayRK5array", false]], "operator^ (c++ function)": [[0, "_CPPv4eoRK5arrayRK5array", false]], "operator| (c++ function)": [[0, "_CPPv4orRK5arrayRK5array", false]], "operator|| (c++ function)": [[0, "_CPPv4ooRK5arrayRK5array", false]], "optimizer (class in mlx.optimizers)": [[474, "mlx.optimizers.Optimizer", false]], "outer (c++ function)": [[0, "_CPPv45outerRK5arrayRK5array14StreamOrDevice", false]], "outer() (in module mlx.core)": [[231, "mlx.core.outer", false]], "pad (c++ function)": [[0, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", false], [0, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", false], [0, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", false], [0, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", false]], "pad() (in module mlx.core)": [[232, "mlx.core.pad", false]], "parameters() (module method)": [[368, "mlx.nn.Module.parameters", false]], "partition (c++ function)": [[0, "_CPPv49partitionRK5arrayi14StreamOrDevice", false], [0, "_CPPv49partitionRK5arrayii14StreamOrDevice", false]], "partition() (in module mlx.core)": [[233, "mlx.core.partition", false]], "permutation() (in module mlx.core.random)": [[247, "mlx.core.random.permutation", false]], "power (c++ function)": [[0, "_CPPv45powerRK5arrayRK5array14StreamOrDevice", false]], "power() (in module mlx.core)": [[234, "mlx.core.power", false]], "prelu (class in mlx.nn)": [[379, "mlx.nn.PReLU", false], [436, "mlx.nn.prelu", false]], "prod (c++ function)": [[0, "_CPPv44prodRK5array14StreamOrDevice", false], [0, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv44prodRK5arrayb14StreamOrDevice", false], [0, "_CPPv44prodRK5arrayib14StreamOrDevice", false]], "prod() (array method)": [[63, "mlx.core.array.prod", false]], "prod() (in module mlx.core)": [[235, "mlx.core.prod", false]], "put_along_axis (c++ function)": [[0, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false]], "put_along_axis() (in module mlx.core)": [[236, "mlx.core.put_along_axis", false]], "qr() (in module mlx.core.linalg)": [[190, "mlx.core.linalg.qr", false]], "quantize (c++ function)": [[0, "_CPPv48quantizeRK5arrayii14StreamOrDevice", false]], "quantize() (in module mlx.core)": [[237, "mlx.core.quantize", false]], "quantize() (in module mlx.nn)": [[307, "mlx.nn.quantize", false]], "quantized_matmul (c++ function)": [[0, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", false]], "quantized_matmul() (in module mlx.core)": [[238, "mlx.core.quantized_matmul", false]], "quantizedembedding (class in mlx.nn)": [[380, "mlx.nn.QuantizedEmbedding", false]], "quantizedlinear (class in mlx.nn)": [[381, "mlx.nn.QuantizedLinear", false]], "radians (c++ function)": [[0, "_CPPv47radiansRK5array14StreamOrDevice", false]], "radians() (in module mlx.core)": [[239, "mlx.core.radians", false]], "randint() (in module mlx.core.random)": [[248, "mlx.core.random.randint", false]], "real (c++ function)": [[0, "_CPPv44realRK5array14StreamOrDevice", false]], "real() (in module mlx.core)": [[253, "mlx.core.real", false]], "reciprocal (c++ function)": [[0, "_CPPv410reciprocalRK5array14StreamOrDevice", false]], "reciprocal() (array method)": [[64, "mlx.core.array.reciprocal", false]], "reciprocal() (in module mlx.core)": [[254, "mlx.core.reciprocal", false]], "recv() (in module mlx.core.distributed)": [[125, "mlx.core.distributed.recv", false]], "recv_like() (in module mlx.core.distributed)": [[126, "mlx.core.distributed.recv_like", false]], "relu (class in mlx.nn)": [[384, "mlx.nn.ReLU", false], [437, "mlx.nn.relu", false]], "relu6 (class in mlx.nn)": [[385, "mlx.nn.ReLU6", false], [438, "mlx.nn.relu6", false]], "remainder (c++ function)": [[0, "_CPPv49remainderRK5arrayRK5array14StreamOrDevice", false]], "remainder() (in module mlx.core)": [[255, "mlx.core.remainder", false]], "repeat (c++ function)": [[0, "_CPPv46repeatRK5arrayi14StreamOrDevice", false], [0, "_CPPv46repeatRK5arrayii14StreamOrDevice", false]], "repeat() (in module mlx.core)": [[256, "mlx.core.repeat", false]], "reset_peak_memory() (in module mlx.core.metal)": [[215, "mlx.core.metal.reset_peak_memory", false]], "reshape (c++ function)": [[0, "_CPPv47reshapeRK5arrayNSt6vectorIiEE14StreamOrDevice", false]], "reshape() (array method)": [[65, "mlx.core.array.reshape", false]], "reshape() (in module mlx.core)": [[257, "mlx.core.reshape", false]], "rfft() (in module mlx.core.fft)": [[156, "mlx.core.fft.rfft", false]], "rfft2() (in module mlx.core.fft)": [[157, "mlx.core.fft.rfft2", false]], "rfftn() (in module mlx.core.fft)": [[158, "mlx.core.fft.rfftn", false]], "right_shift (c++ function)": [[0, "_CPPv411right_shiftRK5arrayRK5array14StreamOrDevice", false]], "right_shift() (in module mlx.core)": [[258, "mlx.core.right_shift", false]], "rms_norm() (in module mlx.core.fast)": [[144, "mlx.core.fast.rms_norm", false]], "rmsnorm (class in mlx.nn)": [[382, "mlx.nn.RMSNorm", false]], "rmsprop (class in mlx.optimizers)": [[466, "mlx.optimizers.RMSprop", false]], "rnn (class in mlx.nn)": [[383, "mlx.nn.RNN", false]], "roll (c++ function)": [[0, "_CPPv44rollRK5arrayRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", false], [0, "_CPPv44rollRK5arrayi14StreamOrDevice", false], [0, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv44rollRK5arrayii14StreamOrDevice", false]], "roll() (in module mlx.core)": [[259, "mlx.core.roll", false]], "rope (class in mlx.nn)": [[386, "mlx.nn.RoPE", false]], "rope() (in module mlx.core.fast)": [[145, "mlx.core.fast.rope", false]], "round (c++ function)": [[0, "_CPPv45roundRK5array14StreamOrDevice", false], [0, "_CPPv45roundRK5arrayi14StreamOrDevice", false]], "round() (array method)": [[66, "mlx.core.array.round", false]], "round() (in module mlx.core)": [[260, "mlx.core.round", false]], "rsqrt (c++ function)": [[0, "_CPPv45rsqrtRK5array14StreamOrDevice", false]], "rsqrt() (array method)": [[67, "mlx.core.array.rsqrt", false]], "rsqrt() (in module mlx.core)": [[261, "mlx.core.rsqrt", false]], "save() (in module mlx.core)": [[262, "mlx.core.save", false]], "save_gguf() (in module mlx.core)": [[263, "mlx.core.save_gguf", false]], "save_safetensors() (in module mlx.core)": [[264, "mlx.core.save_safetensors", false]], "save_weights() (module method)": [[369, "mlx.nn.Module.save_weights", false]], "savez() (in module mlx.core)": [[265, "mlx.core.savez", false]], "savez_compressed() (in module mlx.core)": [[266, "mlx.core.savez_compressed", false]], "scaled_dot_product_attention() (in module mlx.core.fast)": [[146, "mlx.core.fast.scaled_dot_product_attention", false]], "scatter (c++ function)": [[0, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "scatter_add (c++ function)": [[0, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "scatter_max (c++ function)": [[0, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "scatter_min (c++ function)": [[0, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "scatter_prod (c++ function)": [[0, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", false]], "seed() (in module mlx.core.random)": [[249, "mlx.core.random.seed", false]], "selu (class in mlx.nn)": [[387, "mlx.nn.SELU", false], [439, "mlx.nn.selu", false]], "send() (in module mlx.core.distributed)": [[127, "mlx.core.distributed.send", false]], "sequential (class in mlx.nn)": [[388, "mlx.nn.Sequential", false]], "set_cache_limit() (in module mlx.core.metal)": [[216, "mlx.core.metal.set_cache_limit", false]], "set_default_device() (in module mlx.core)": [[267, "mlx.core.set_default_device", false]], "set_default_stream() (in module mlx.core)": [[268, "mlx.core.set_default_stream", false]], "set_dtype() (module method)": [[370, "mlx.nn.Module.set_dtype", false]], "set_memory_limit() (in module mlx.core.metal)": [[217, "mlx.core.metal.set_memory_limit", false]], "set_wired_limit() (in module mlx.core.metal)": [[218, "mlx.core.metal.set_wired_limit", false]], "sgd (class in mlx.optimizers)": [[467, "mlx.optimizers.SGD", false]], "shape (array property)": [[68, "mlx.core.array.shape", false]], "sigmoid (c++ function)": [[0, "_CPPv47sigmoidRK5array14StreamOrDevice", false]], "sigmoid (class in mlx.nn)": [[390, "mlx.nn.Sigmoid", false], [440, "mlx.nn.sigmoid", false]], "sigmoid() (in module mlx.core)": [[269, "mlx.core.sigmoid", false]], "sign (c++ function)": [[0, "_CPPv44signRK5array14StreamOrDevice", false]], "sign() (in module mlx.core)": [[270, "mlx.core.sign", false]], "silu (class in mlx.nn)": [[389, "mlx.nn.SiLU", false], [441, "mlx.nn.silu", false]], "sin (c++ function)": [[0, "_CPPv43sinRK5array14StreamOrDevice", false]], "sin() (array method)": [[69, "mlx.core.array.sin", false]], "sin() (in module mlx.core)": [[271, "mlx.core.sin", false]], "sinh (c++ function)": [[0, "_CPPv44sinhRK5array14StreamOrDevice", false]], "sinh() (in module mlx.core)": [[272, "mlx.core.sinh", false]], "sinusoidalpositionalencoding (class in mlx.nn)": [[391, "mlx.nn.SinusoidalPositionalEncoding", false]], "size (array property)": [[70, "mlx.core.array.size", false]], "slice (c++ function)": [[0, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", false]], "slice_update (c++ function)": [[0, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", false]], "smooth_l1_loss (class in mlx.nn.losses)": [[433, "mlx.nn.losses.smooth_l1_loss", false]], "softmax (c++ function)": [[0, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv47softmaxRK5arrayb14StreamOrDevice", false], [0, "_CPPv47softmaxRK5arrayib14StreamOrDevice", false]], "softmax (class in mlx.nn)": [[392, "mlx.nn.Softmax", false], [442, "mlx.nn.softmax", false]], "softmax() (in module mlx.core)": [[273, "mlx.core.softmax", false]], "softmin (class in mlx.nn)": [[393, "mlx.nn.Softmin", false], [443, "mlx.nn.softmin", false]], "softplus (class in mlx.nn)": [[394, "mlx.nn.Softplus", false], [444, "mlx.nn.softplus", false]], "softshrink (class in mlx.nn)": [[395, "mlx.nn.Softshrink", false], [445, "mlx.nn.softshrink", false]], "softsign (class in mlx.nn)": [[396, "mlx.nn.Softsign", false]], "sort (c++ function)": [[0, "_CPPv44sortRK5array14StreamOrDevice", false], [0, "_CPPv44sortRK5arrayi14StreamOrDevice", false]], "sort() (in module mlx.core)": [[274, "mlx.core.sort", false]], "split (c++ function)": [[0, "_CPPv45splitRK5arrayRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", false], [0, "_CPPv45splitRK5arrayi14StreamOrDevice", false], [0, "_CPPv45splitRK5arrayii14StreamOrDevice", false]], "split() (array method)": [[71, "mlx.core.array.split", false]], "split() (in module mlx.core)": [[275, "mlx.core.split", false]], "split() (in module mlx.core.random)": [[250, "mlx.core.random.split", false]], "sqrt (c++ function)": [[0, "_CPPv44sqrtRK5array14StreamOrDevice", false]], "sqrt() (array method)": [[72, "mlx.core.array.sqrt", false]], "sqrt() (in module mlx.core)": [[276, "mlx.core.sqrt", false]], "square (c++ function)": [[0, "_CPPv46squareRK5array14StreamOrDevice", false]], "square() (array method)": [[73, "mlx.core.array.square", false]], "square() (in module mlx.core)": [[277, "mlx.core.square", false]], "squeeze (c++ function)": [[0, "_CPPv47squeezeRK5array14StreamOrDevice", false], [0, "_CPPv47squeezeRK5arrayRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv47squeezeRK5arrayi14StreamOrDevice", false]], "squeeze() (array method)": [[74, "mlx.core.array.squeeze", false]], "squeeze() (in module mlx.core)": [[278, "mlx.core.squeeze", false]], "stack (c++ function)": [[0, "_CPPv45stackRKNSt6vectorI5arrayEE14StreamOrDevice", false], [0, "_CPPv45stackRKNSt6vectorI5arrayEEi14StreamOrDevice", false]], "stack() (in module mlx.core)": [[279, "mlx.core.stack", false]], "start_capture() (in module mlx.core.metal)": [[219, "mlx.core.metal.start_capture", false]], "state (module property)": [[371, "mlx.nn.Module.state", false]], "state (optimizer property)": [[464, "mlx.optimizers.Optimizer.state", false]], "std (c++ function)": [[0, "_CPPv4StRK5array14StreamOrDevice", false], [0, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", false], [0, "_CPPv4StRK5arraybi14StreamOrDevice", false], [0, "_CPPv4StRK5arrayibi14StreamOrDevice", false]], "std() (array method)": [[75, "mlx.core.array.std", false]], "std() (in module mlx.core)": [[280, "mlx.core.std", false]], "step (class in mlx.nn)": [[397, "mlx.nn.Step", false], [446, "mlx.nn.step", false]], "step_decay() (in module mlx.optimizers)": [[472, "mlx.optimizers.step_decay", false]], "stop_capture() (in module mlx.core.metal)": [[220, "mlx.core.metal.stop_capture", false]], "stop_gradient (c++ function)": [[0, "_CPPv413stop_gradientRK5array14StreamOrDevice", false]], "stop_gradient() (in module mlx.core)": [[281, "mlx.core.stop_gradient", false]], "stream (class in mlx.core)": [[315, "mlx.core.Stream", false]], "stream() (in module mlx.core)": [[282, "mlx.core.stream", false]], "subtract (c++ function)": [[0, "_CPPv48subtractRK5arrayRK5array14StreamOrDevice", false]], "subtract() (in module mlx.core)": [[283, "mlx.core.subtract", false]], "sum (c++ function)": [[0, "_CPPv43sumRK5array14StreamOrDevice", false], [0, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", false], [0, "_CPPv43sumRK5arrayb14StreamOrDevice", false], [0, "_CPPv43sumRK5arrayib14StreamOrDevice", false]], "sum() (array method)": [[76, "mlx.core.array.sum", false]], "sum() (in module mlx.core)": [[284, "mlx.core.sum", false]], "svd() (in module mlx.core.linalg)": [[191, "mlx.core.linalg.svd", false]], "swapaxes (c++ function)": [[0, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", false]], "swapaxes() (array method)": [[77, "mlx.core.array.swapaxes", false]], "swapaxes() (in module mlx.core)": [[285, "mlx.core.swapaxes", false]], "synchronize() (in module mlx.core)": [[286, "mlx.core.synchronize", false]], "t (array property)": [[31, "mlx.core.array.T", false]], "take (c++ function)": [[0, "_CPPv44takeRK5arrayRK5array14StreamOrDevice", false], [0, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", false], [0, "_CPPv44takeRK5arrayi14StreamOrDevice", false], [0, "_CPPv44takeRK5arrayii14StreamOrDevice", false]], "take() (in module mlx.core)": [[287, "mlx.core.take", false]], "take_along_axis (c++ function)": [[0, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", false]], "take_along_axis() (in module mlx.core)": [[288, "mlx.core.take_along_axis", false]], "tan (c++ function)": [[0, "_CPPv43tanRK5array14StreamOrDevice", false]], "tan() (in module mlx.core)": [[289, "mlx.core.tan", false]], "tanh (c++ function)": [[0, "_CPPv44tanhRK5array14StreamOrDevice", false]], "tanh (class in mlx.nn)": [[398, "mlx.nn.Tanh", false], [447, "mlx.nn.tanh", false]], "tanh() (in module mlx.core)": [[290, "mlx.core.tanh", false]], "tensordot (c++ function)": [[0, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", false], [0, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", false]], "tensordot() (in module mlx.core)": [[291, "mlx.core.tensordot", false]], "tile (c++ function)": [[0, "_CPPv44tileRK5arrayNSt6vectorIiEE14StreamOrDevice", false]], "tile() (in module mlx.core)": [[292, "mlx.core.tile", false]], "tolist() (array method)": [[78, "mlx.core.array.tolist", false]], "topk (c++ function)": [[0, "_CPPv44topkRK5arrayi14StreamOrDevice", false], [0, "_CPPv44topkRK5arrayii14StreamOrDevice", false]], "topk() (in module mlx.core)": [[293, "mlx.core.topk", false]], "trace (c++ function)": [[0, "_CPPv45traceRK5array14StreamOrDevice", false], [0, "_CPPv45traceRK5arrayiii14StreamOrDevice", false], [0, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", false]], "trace() (in module mlx.core)": [[294, "mlx.core.trace", false]], "train() (module method)": [[372, "mlx.nn.Module.train", false]], "trainable_parameters() (module method)": [[373, "mlx.nn.Module.trainable_parameters", false]], "training (module property)": [[374, "mlx.nn.Module.training", false]], "transformer (class in mlx.nn)": [[399, "mlx.nn.Transformer", false]], "transpose (c++ function)": [[0, "_CPPv49transposeRK5array14StreamOrDevice", false], [0, "_CPPv49transposeRK5arrayNSt16initializer_listIiEE14StreamOrDevice", false], [0, "_CPPv49transposeRK5arrayNSt6vectorIiEE14StreamOrDevice", false]], "transpose() (array method)": [[79, "mlx.core.array.transpose", false]], "transpose() (in module mlx.core)": [[295, "mlx.core.transpose", false]], "tree_flatten() (in module mlx.utils)": [[310, "mlx.utils.tree_flatten", false]], "tree_map() (in module mlx.utils)": [[311, "mlx.utils.tree_map", false]], "tree_map_with_path() (in module mlx.utils)": [[312, "mlx.utils.tree_map_with_path", false]], "tree_reduce() (in module mlx.utils)": [[313, "mlx.utils.tree_reduce", false]], "tree_unflatten() (in module mlx.utils)": [[314, "mlx.utils.tree_unflatten", false]], "tri (c++ function)": [[0, "_CPPv43trii5Dtype14StreamOrDevice", false], [0, "_CPPv43triiii5Dtype14StreamOrDevice", false]], "tri() (in module mlx.core)": [[296, "mlx.core.tri", false]], "tri_inv() (in module mlx.core.linalg)": [[192, "mlx.core.linalg.tri_inv", false]], "tril (c++ function)": [[0, "_CPPv44tril5arrayi14StreamOrDevice", false]], "tril() (in module mlx.core)": [[297, "mlx.core.tril", false]], "triplet_loss (class in mlx.nn.losses)": [[434, "mlx.nn.losses.triplet_loss", false]], "triu (c++ function)": [[0, "_CPPv44triu5arrayi14StreamOrDevice", false]], "triu() (in module mlx.core)": [[298, "mlx.core.triu", false]], "truncated_normal() (in module mlx.core.random)": [[251, "mlx.core.random.truncated_normal", false]], "unfreeze() (module method)": [[375, "mlx.nn.Module.unfreeze", false]], "uniform() (in module mlx.core.random)": [[252, "mlx.core.random.uniform", false]], "uniform() (in module mlx.nn.init)": [[408, "mlx.nn.init.uniform", false]], "update() (module method)": [[376, "mlx.nn.Module.update", false]], "update() (optimizer method)": [[465, "mlx.optimizers.Optimizer.update", false]], "update_modules() (module method)": [[377, "mlx.nn.Module.update_modules", false]], "upsample (class in mlx.nn)": [[400, "mlx.nn.Upsample", false]], "value_and_grad() (in module mlx.core)": [[299, "mlx.core.value_and_grad", false]], "value_and_grad() (in module mlx.nn)": [[308, "mlx.nn.value_and_grad", false]], "var (c++ function)": [[0, "_CPPv43varRK5array14StreamOrDevice", false], [0, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", false], [0, "_CPPv43varRK5arraybi14StreamOrDevice", false], [0, "_CPPv43varRK5arrayibi14StreamOrDevice", false]], "var() (array method)": [[80, "mlx.core.array.var", false]], "var() (in module mlx.core)": [[300, "mlx.core.var", false]], "view (c++ function)": [[0, "_CPPv44viewRK5arrayRK5Dtype14StreamOrDevice", false]], "view() (array method)": [[81, "mlx.core.array.view", false]], "view() (in module mlx.core)": [[301, "mlx.core.view", false]], "vjp() (in module mlx.core)": [[302, "mlx.core.vjp", false]], "vmap() (in module mlx.core)": [[303, "mlx.core.vmap", false]], "where (c++ function)": [[0, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", false]], "where() (in module mlx.core)": [[304, "mlx.core.where", false]], "zeros (c++ function)": [[0, "_CPPv45zerosRKNSt6vectorIiEE14StreamOrDevice", false], [0, "_CPPv45zerosRKNSt6vectorIiEE5Dtype14StreamOrDevice", false]], "zeros() (in module mlx.core)": [[305, "mlx.core.zeros", false]], "zeros_like (c++ function)": [[0, "_CPPv410zeros_likeRK5array14StreamOrDevice", false]], "zeros_like() (in module mlx.core)": [[306, "mlx.core.zeros_like", false]]}, "objects": {"": [[0, 0, 1, "_CPPv43absRK5array14StreamOrDevice", "abs"], [0, 1, 1, "_CPPv43absRK5array14StreamOrDevice", "abs::a"], [0, 1, 1, "_CPPv43absRK5array14StreamOrDevice", "abs::s"], [0, 0, 1, "_CPPv43addRK5arrayRK5array14StreamOrDevice", "add"], [0, 1, 1, "_CPPv43addRK5arrayRK5array14StreamOrDevice", "add::a"], [0, 1, 1, "_CPPv43addRK5arrayRK5array14StreamOrDevice", "add::b"], [0, 1, 1, "_CPPv43addRK5arrayRK5array14StreamOrDevice", "add::s"], [0, 0, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::a"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::alpha"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::b"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::beta"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::c"], [0, 1, 1, "_CPPv45addmm5array5array5arrayRKfRKf14StreamOrDevice", "addmm::s"], [0, 0, 1, "_CPPv43allRK5array14StreamOrDevice", "all"], [0, 0, 1, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "all"], [0, 0, 1, "_CPPv43allRK5arrayb14StreamOrDevice", "all"], [0, 0, 1, "_CPPv43allRK5arrayib14StreamOrDevice", "all"], [0, 1, 1, "_CPPv43allRK5array14StreamOrDevice", "all::a"], [0, 1, 1, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "all::a"], [0, 1, 1, "_CPPv43allRK5arrayb14StreamOrDevice", "all::a"], [0, 1, 1, "_CPPv43allRK5arrayib14StreamOrDevice", "all::a"], [0, 1, 1, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "all::axes"], [0, 1, 1, "_CPPv43allRK5arrayib14StreamOrDevice", "all::axis"], [0, 1, 1, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "all::keepdims"], [0, 1, 1, "_CPPv43allRK5arrayb14StreamOrDevice", "all::keepdims"], [0, 1, 1, "_CPPv43allRK5arrayib14StreamOrDevice", "all::keepdims"], [0, 1, 1, "_CPPv43allRK5array14StreamOrDevice", "all::s"], [0, 1, 1, "_CPPv43allRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "all::s"], [0, 1, 1, "_CPPv43allRK5arrayb14StreamOrDevice", "all::s"], [0, 1, 1, "_CPPv43allRK5arrayib14StreamOrDevice", "all::s"], [0, 0, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::a"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::atol"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::b"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::equal_nan"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::rtol"], [0, 1, 1, "_CPPv48allcloseRK5arrayRK5arrayddb14StreamOrDevice", "allclose::s"], [0, 0, 1, "_CPPv43anyRK5array14StreamOrDevice", "any"], [0, 0, 1, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "any"], [0, 0, 1, "_CPPv43anyRK5arrayb14StreamOrDevice", "any"], [0, 0, 1, "_CPPv43anyRK5arrayib14StreamOrDevice", "any"], [0, 1, 1, "_CPPv43anyRK5array14StreamOrDevice", "any::a"], [0, 1, 1, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "any::a"], [0, 1, 1, "_CPPv43anyRK5arrayb14StreamOrDevice", "any::a"], [0, 1, 1, "_CPPv43anyRK5arrayib14StreamOrDevice", "any::a"], [0, 1, 1, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "any::axes"], [0, 1, 1, "_CPPv43anyRK5arrayib14StreamOrDevice", "any::axis"], [0, 1, 1, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "any::keepdims"], [0, 1, 1, "_CPPv43anyRK5arrayb14StreamOrDevice", "any::keepdims"], [0, 1, 1, "_CPPv43anyRK5arrayib14StreamOrDevice", "any::keepdims"], [0, 1, 1, "_CPPv43anyRK5array14StreamOrDevice", "any::s"], [0, 1, 1, "_CPPv43anyRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "any::s"], [0, 1, 1, "_CPPv43anyRK5arrayb14StreamOrDevice", "any::s"], [0, 1, 1, "_CPPv43anyRK5arrayib14StreamOrDevice", "any::s"], [0, 0, 1, "_CPPv46aranged14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46aranged5Dtype14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangedd14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangedd5Dtype14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangeddd14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangei14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangeii14StreamOrDevice", "arange"], [0, 0, 1, "_CPPv46arangeiii14StreamOrDevice", "arange"], [0, 1, 1, "_CPPv46aranged5Dtype14StreamOrDevice", "arange::dtype"], [0, 1, 1, "_CPPv46arangedd5Dtype14StreamOrDevice", "arange::dtype"], [0, 1, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange::dtype"], [0, 1, 1, "_CPPv46aranged14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46aranged5Dtype14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangedd14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangedd5Dtype14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangeddd14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangei14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangeii14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangeiii14StreamOrDevice", "arange::s"], [0, 1, 1, "_CPPv46arangedd14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangedd5Dtype14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangeddd14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangeii14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangeiii14StreamOrDevice", "arange::start"], [0, 1, 1, "_CPPv46arangeddd14StreamOrDevice", "arange::step"], [0, 1, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange::step"], [0, 1, 1, "_CPPv46arangeiii14StreamOrDevice", "arange::step"], [0, 1, 1, "_CPPv46aranged14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46aranged5Dtype14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangedd14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangedd5Dtype14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangeddd14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangeddd5Dtype14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangei14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangeii14StreamOrDevice", "arange::stop"], [0, 1, 1, "_CPPv46arangeiii14StreamOrDevice", "arange::stop"], [0, 0, 1, "_CPPv46arccosRK5array14StreamOrDevice", "arccos"], [0, 1, 1, "_CPPv46arccosRK5array14StreamOrDevice", "arccos::a"], [0, 1, 1, "_CPPv46arccosRK5array14StreamOrDevice", "arccos::s"], [0, 0, 1, "_CPPv47arccoshRK5array14StreamOrDevice", "arccosh"], [0, 1, 1, "_CPPv47arccoshRK5array14StreamOrDevice", "arccosh::a"], [0, 1, 1, "_CPPv47arccoshRK5array14StreamOrDevice", "arccosh::s"], [0, 0, 1, "_CPPv46arcsinRK5array14StreamOrDevice", "arcsin"], [0, 1, 1, "_CPPv46arcsinRK5array14StreamOrDevice", "arcsin::a"], [0, 1, 1, "_CPPv46arcsinRK5array14StreamOrDevice", "arcsin::s"], [0, 0, 1, "_CPPv47arcsinhRK5array14StreamOrDevice", "arcsinh"], [0, 1, 1, "_CPPv47arcsinhRK5array14StreamOrDevice", "arcsinh::a"], [0, 1, 1, "_CPPv47arcsinhRK5array14StreamOrDevice", "arcsinh::s"], [0, 0, 1, "_CPPv46arctanRK5array14StreamOrDevice", "arctan"], [0, 0, 1, "_CPPv47arctan2RK5arrayRK5array14StreamOrDevice", "arctan2"], [0, 1, 1, "_CPPv47arctan2RK5arrayRK5array14StreamOrDevice", "arctan2::a"], [0, 1, 1, "_CPPv47arctan2RK5arrayRK5array14StreamOrDevice", "arctan2::b"], [0, 1, 1, "_CPPv47arctan2RK5arrayRK5array14StreamOrDevice", "arctan2::s"], [0, 1, 1, "_CPPv46arctanRK5array14StreamOrDevice", "arctan::a"], [0, 1, 1, "_CPPv46arctanRK5array14StreamOrDevice", "arctan::s"], [0, 0, 1, "_CPPv47arctanhRK5array14StreamOrDevice", "arctanh"], [0, 1, 1, "_CPPv47arctanhRK5array14StreamOrDevice", "arctanh::a"], [0, 1, 1, "_CPPv47arctanhRK5array14StreamOrDevice", "arctanh::s"], [0, 0, 1, "_CPPv46argmaxRK5array14StreamOrDevice", "argmax"], [0, 0, 1, "_CPPv46argmaxRK5arrayb14StreamOrDevice", "argmax"], [0, 0, 1, "_CPPv46argmaxRK5arrayib14StreamOrDevice", "argmax"], [0, 1, 1, "_CPPv46argmaxRK5array14StreamOrDevice", "argmax::a"], [0, 1, 1, "_CPPv46argmaxRK5arrayb14StreamOrDevice", "argmax::a"], [0, 1, 1, "_CPPv46argmaxRK5arrayib14StreamOrDevice", "argmax::a"], [0, 1, 1, "_CPPv46argmaxRK5arrayib14StreamOrDevice", "argmax::axis"], [0, 1, 1, "_CPPv46argmaxRK5arrayb14StreamOrDevice", "argmax::keepdims"], [0, 1, 1, "_CPPv46argmaxRK5arrayib14StreamOrDevice", "argmax::keepdims"], [0, 1, 1, "_CPPv46argmaxRK5array14StreamOrDevice", "argmax::s"], [0, 1, 1, "_CPPv46argmaxRK5arrayb14StreamOrDevice", "argmax::s"], [0, 1, 1, "_CPPv46argmaxRK5arrayib14StreamOrDevice", "argmax::s"], [0, 0, 1, "_CPPv46argminRK5array14StreamOrDevice", "argmin"], [0, 0, 1, "_CPPv46argminRK5arrayb14StreamOrDevice", "argmin"], [0, 0, 1, "_CPPv46argminRK5arrayib14StreamOrDevice", "argmin"], [0, 1, 1, "_CPPv46argminRK5array14StreamOrDevice", "argmin::a"], [0, 1, 1, "_CPPv46argminRK5arrayb14StreamOrDevice", "argmin::a"], [0, 1, 1, "_CPPv46argminRK5arrayib14StreamOrDevice", "argmin::a"], [0, 1, 1, "_CPPv46argminRK5arrayib14StreamOrDevice", "argmin::axis"], [0, 1, 1, "_CPPv46argminRK5arrayb14StreamOrDevice", "argmin::keepdims"], [0, 1, 1, "_CPPv46argminRK5arrayib14StreamOrDevice", "argmin::keepdims"], [0, 1, 1, "_CPPv46argminRK5array14StreamOrDevice", "argmin::s"], [0, 1, 1, "_CPPv46argminRK5arrayb14StreamOrDevice", "argmin::s"], [0, 1, 1, "_CPPv46argminRK5arrayib14StreamOrDevice", "argmin::s"], [0, 0, 1, "_CPPv412argpartitionRK5arrayi14StreamOrDevice", "argpartition"], [0, 0, 1, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", "argpartition"], [0, 1, 1, "_CPPv412argpartitionRK5arrayi14StreamOrDevice", "argpartition::a"], [0, 1, 1, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", "argpartition::a"], [0, 1, 1, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", "argpartition::axis"], [0, 1, 1, "_CPPv412argpartitionRK5arrayi14StreamOrDevice", "argpartition::kth"], [0, 1, 1, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", "argpartition::kth"], [0, 1, 1, "_CPPv412argpartitionRK5arrayi14StreamOrDevice", "argpartition::s"], [0, 1, 1, "_CPPv412argpartitionRK5arrayii14StreamOrDevice", "argpartition::s"], [0, 0, 1, "_CPPv47argsortRK5array14StreamOrDevice", "argsort"], [0, 0, 1, "_CPPv47argsortRK5arrayi14StreamOrDevice", "argsort"], [0, 1, 1, "_CPPv47argsortRK5array14StreamOrDevice", "argsort::a"], [0, 1, 1, "_CPPv47argsortRK5arrayi14StreamOrDevice", "argsort::a"], [0, 1, 1, "_CPPv47argsortRK5arrayi14StreamOrDevice", "argsort::axis"], [0, 1, 1, "_CPPv47argsortRK5array14StreamOrDevice", "argsort::s"], [0, 1, 1, "_CPPv47argsortRK5arrayi14StreamOrDevice", "argsort::s"], [0, 0, 1, "_CPPv411array_equalRK5arrayRK5array14StreamOrDevice", "array_equal"], [0, 0, 1, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", "array_equal"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5array14StreamOrDevice", "array_equal::a"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", "array_equal::a"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5array14StreamOrDevice", "array_equal::b"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", "array_equal::b"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", "array_equal::equal_nan"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5array14StreamOrDevice", "array_equal::s"], [0, 1, 1, "_CPPv411array_equalRK5arrayRK5arrayb14StreamOrDevice", "array_equal::s"], [0, 0, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided"], [0, 1, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided::a"], [0, 1, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided::offset"], [0, 1, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided::s"], [0, 1, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided::shape"], [0, 1, 1, "_CPPv410as_strided5arrayNSt6vectorIiEENSt6vectorI6size_tEE6size_t14StreamOrDevice", "as_strided::strides"], [0, 0, 1, "_CPPv46astype5array5Dtype14StreamOrDevice", "astype"], [0, 1, 1, "_CPPv46astype5array5Dtype14StreamOrDevice", "astype::a"], [0, 1, 1, "_CPPv46astype5array5Dtype14StreamOrDevice", "astype::dtype"], [0, 1, 1, "_CPPv46astype5array5Dtype14StreamOrDevice", "astype::s"], [0, 0, 1, "_CPPv410atleast_1dRK5array14StreamOrDevice", "atleast_1d"], [0, 0, 1, "_CPPv410atleast_1dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_1d"], [0, 1, 1, "_CPPv410atleast_1dRK5array14StreamOrDevice", "atleast_1d::a"], [0, 1, 1, "_CPPv410atleast_1dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_1d::a"], [0, 1, 1, "_CPPv410atleast_1dRK5array14StreamOrDevice", "atleast_1d::s"], [0, 1, 1, "_CPPv410atleast_1dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_1d::s"], [0, 0, 1, "_CPPv410atleast_2dRK5array14StreamOrDevice", "atleast_2d"], [0, 0, 1, "_CPPv410atleast_2dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_2d"], [0, 1, 1, "_CPPv410atleast_2dRK5array14StreamOrDevice", "atleast_2d::a"], [0, 1, 1, "_CPPv410atleast_2dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_2d::a"], [0, 1, 1, "_CPPv410atleast_2dRK5array14StreamOrDevice", "atleast_2d::s"], [0, 1, 1, "_CPPv410atleast_2dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_2d::s"], [0, 0, 1, "_CPPv410atleast_3dRK5array14StreamOrDevice", "atleast_3d"], [0, 0, 1, "_CPPv410atleast_3dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_3d"], [0, 1, 1, "_CPPv410atleast_3dRK5array14StreamOrDevice", "atleast_3d::a"], [0, 1, 1, "_CPPv410atleast_3dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_3d::a"], [0, 1, 1, "_CPPv410atleast_3dRK5array14StreamOrDevice", "atleast_3d::s"], [0, 1, 1, "_CPPv410atleast_3dRKNSt6vectorI5arrayEE14StreamOrDevice", "atleast_3d::s"], [0, 0, 1, "_CPPv411bitwise_andRK5arrayRK5array14StreamOrDevice", "bitwise_and"], [0, 1, 1, "_CPPv411bitwise_andRK5arrayRK5array14StreamOrDevice", "bitwise_and::a"], [0, 1, 1, "_CPPv411bitwise_andRK5arrayRK5array14StreamOrDevice", "bitwise_and::b"], [0, 1, 1, "_CPPv411bitwise_andRK5arrayRK5array14StreamOrDevice", "bitwise_and::s"], [0, 0, 1, "_CPPv410bitwise_orRK5arrayRK5array14StreamOrDevice", "bitwise_or"], [0, 1, 1, "_CPPv410bitwise_orRK5arrayRK5array14StreamOrDevice", "bitwise_or::a"], [0, 1, 1, "_CPPv410bitwise_orRK5arrayRK5array14StreamOrDevice", "bitwise_or::b"], [0, 1, 1, "_CPPv410bitwise_orRK5arrayRK5array14StreamOrDevice", "bitwise_or::s"], [0, 0, 1, "_CPPv411bitwise_xorRK5arrayRK5array14StreamOrDevice", "bitwise_xor"], [0, 1, 1, "_CPPv411bitwise_xorRK5arrayRK5array14StreamOrDevice", "bitwise_xor::a"], [0, 1, 1, "_CPPv411bitwise_xorRK5arrayRK5array14StreamOrDevice", "bitwise_xor::b"], [0, 1, 1, "_CPPv411bitwise_xorRK5arrayRK5array14StreamOrDevice", "bitwise_xor::s"], [0, 0, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::a"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::b"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::block_size"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::mask_lhs"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::mask_out"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::mask_rhs"], [0, 1, 1, "_CPPv415block_masked_mm5array5arrayiNSt8optionalI5arrayEENSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "block_masked_mm::s"], [0, 0, 1, "_CPPv416broadcast_arraysRKNSt6vectorI5arrayEE14StreamOrDevice", "broadcast_arrays"], [0, 1, 1, "_CPPv416broadcast_arraysRKNSt6vectorI5arrayEE14StreamOrDevice", "broadcast_arrays::inputs"], [0, 1, 1, "_CPPv416broadcast_arraysRKNSt6vectorI5arrayEE14StreamOrDevice", "broadcast_arrays::s"], [0, 0, 1, "_CPPv412broadcast_toRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "broadcast_to"], [0, 1, 1, "_CPPv412broadcast_toRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "broadcast_to::a"], [0, 1, 1, "_CPPv412broadcast_toRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "broadcast_to::s"], [0, 1, 1, "_CPPv412broadcast_toRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "broadcast_to::shape"], [0, 0, 1, "_CPPv44ceilRK5array14StreamOrDevice", "ceil"], [0, 1, 1, "_CPPv44ceilRK5array14StreamOrDevice", "ceil::a"], [0, 1, 1, "_CPPv44ceilRK5array14StreamOrDevice", "ceil::s"], [0, 0, 1, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", "clip"], [0, 1, 1, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", "clip::a"], [0, 1, 1, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", "clip::a_max"], [0, 1, 1, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", "clip::a_min"], [0, 1, 1, "_CPPv44clipRK5arrayRKNSt8optionalI5arrayEERKNSt8optionalI5arrayEE14StreamOrDevice", "clip::s"], [0, 0, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEE14StreamOrDevice", "concatenate"], [0, 0, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEEi14StreamOrDevice", "concatenate"], [0, 1, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEE14StreamOrDevice", "concatenate::arrays"], [0, 1, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEEi14StreamOrDevice", "concatenate::arrays"], [0, 1, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEEi14StreamOrDevice", "concatenate::axis"], [0, 1, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEE14StreamOrDevice", "concatenate::s"], [0, 1, 1, "_CPPv411concatenateRKNSt6vectorI5arrayEEi14StreamOrDevice", "concatenate::s"], [0, 0, 1, "_CPPv49conjugateRK5array14StreamOrDevice", "conjugate"], [0, 1, 1, "_CPPv49conjugateRK5array14StreamOrDevice", "conjugate::a"], [0, 1, 1, "_CPPv49conjugateRK5array14StreamOrDevice", "conjugate::s"], [0, 0, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::dilation"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::groups"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::input"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::padding"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::s"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::stride"], [0, 1, 1, "_CPPv46conv1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv1d::weight"], [0, 0, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::dilation"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::groups"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::input"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::padding"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::s"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::stride"], [0, 1, 1, "_CPPv46conv2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv2d::weight"], [0, 0, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::dilation"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::groups"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::input"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::padding"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::s"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::stride"], [0, 1, 1, "_CPPv46conv3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv3d::weight"], [0, 0, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general"], [0, 0, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::flip"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::flip"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::groups"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::groups"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::input"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::input"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::input_dilation"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::input_dilation"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::kernel_dilation"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::kernel_dilation"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::padding"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::padding_hi"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::padding_lo"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::s"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::s"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::stride"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::stride"], [0, 1, 1, "_CPPv412conv_general5array5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::weight"], [0, 1, 1, "_CPPv412conv_generalRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEENSt6vectorIiEEib14StreamOrDevice", "conv_general::weight"], [0, 0, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::dilation"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::groups"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::input"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::padding"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::s"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::stride"], [0, 1, 1, "_CPPv416conv_transpose1dRK5arrayRK5arrayiiii14StreamOrDevice", "conv_transpose1d::weight"], [0, 0, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::dilation"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::groups"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::input"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::padding"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::s"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::stride"], [0, 1, 1, "_CPPv416conv_transpose2dRK5arrayRK5arrayRKNSt4pairIiiEERKNSt4pairIiiEERKNSt4pairIiiEEi14StreamOrDevice", "conv_transpose2d::weight"], [0, 0, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::dilation"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::groups"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::input"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::padding"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::s"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::stride"], [0, 1, 1, "_CPPv416conv_transpose3dRK5arrayRK5arrayRKNSt5tupleIiiiEERKNSt5tupleIiiiEERKNSt5tupleIiiiEEi14StreamOrDevice", "conv_transpose3d::weight"], [0, 0, 1, "_CPPv44copy5array14StreamOrDevice", "copy"], [0, 1, 1, "_CPPv44copy5array14StreamOrDevice", "copy::a"], [0, 1, 1, "_CPPv44copy5array14StreamOrDevice", "copy::s"], [0, 0, 1, "_CPPv43cosRK5array14StreamOrDevice", "cos"], [0, 1, 1, "_CPPv43cosRK5array14StreamOrDevice", "cos::a"], [0, 1, 1, "_CPPv43cosRK5array14StreamOrDevice", "cos::s"], [0, 0, 1, "_CPPv44coshRK5array14StreamOrDevice", "cosh"], [0, 1, 1, "_CPPv44coshRK5array14StreamOrDevice", "cosh::a"], [0, 1, 1, "_CPPv44coshRK5array14StreamOrDevice", "cosh::s"], [0, 0, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax"], [0, 1, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax::a"], [0, 1, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax::axis"], [0, 1, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax::inclusive"], [0, 1, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax::reverse"], [0, 1, 1, "_CPPv46cummaxRK5arrayibb14StreamOrDevice", "cummax::s"], [0, 0, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin"], [0, 1, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin::a"], [0, 1, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin::axis"], [0, 1, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin::inclusive"], [0, 1, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin::reverse"], [0, 1, 1, "_CPPv46cumminRK5arrayibb14StreamOrDevice", "cummin::s"], [0, 0, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod"], [0, 1, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod::a"], [0, 1, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod::axis"], [0, 1, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod::inclusive"], [0, 1, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod::reverse"], [0, 1, 1, "_CPPv47cumprodRK5arrayibb14StreamOrDevice", "cumprod::s"], [0, 0, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum"], [0, 1, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum::a"], [0, 1, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum::axis"], [0, 1, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum::inclusive"], [0, 1, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum::reverse"], [0, 1, 1, "_CPPv46cumsumRK5arrayibb14StreamOrDevice", "cumsum::s"], [0, 0, 1, "_CPPv47degreesRK5array14StreamOrDevice", "degrees"], [0, 1, 1, "_CPPv47degreesRK5array14StreamOrDevice", "degrees::a"], [0, 1, 1, "_CPPv47degreesRK5array14StreamOrDevice", "degrees::s"], [0, 0, 1, "_CPPv47dependsRKNSt6vectorI5arrayEERKNSt6vectorI5arrayEE", "depends"], [0, 1, 1, "_CPPv47dependsRKNSt6vectorI5arrayEERKNSt6vectorI5arrayEE", "depends::dependencies"], [0, 1, 1, "_CPPv47dependsRKNSt6vectorI5arrayEERKNSt6vectorI5arrayEE", "depends::inputs"], [0, 0, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::biases"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::bits"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::group_size"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::s"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::scales"], [0, 1, 1, "_CPPv410dequantizeRK5arrayRK5arrayRK5arrayii14StreamOrDevice", "dequantize::w"], [0, 0, 1, "_CPPv44diagRK5arrayi14StreamOrDevice", "diag"], [0, 1, 1, "_CPPv44diagRK5arrayi14StreamOrDevice", "diag::a"], [0, 1, 1, "_CPPv44diagRK5arrayi14StreamOrDevice", "diag::k"], [0, 1, 1, "_CPPv44diagRK5arrayi14StreamOrDevice", "diag::s"], [0, 0, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal"], [0, 1, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal::a"], [0, 1, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal::axis1"], [0, 1, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal::axis2"], [0, 1, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal::offset"], [0, 1, 1, "_CPPv48diagonalRK5arrayiii14StreamOrDevice", "diagonal::s"], [0, 0, 1, "_CPPv46divideRK5arrayRK5array14StreamOrDevice", "divide"], [0, 1, 1, "_CPPv46divideRK5arrayRK5array14StreamOrDevice", "divide::a"], [0, 1, 1, "_CPPv46divideRK5arrayRK5array14StreamOrDevice", "divide::b"], [0, 1, 1, "_CPPv46divideRK5arrayRK5array14StreamOrDevice", "divide::s"], [0, 0, 1, "_CPPv46divmodRK5arrayRK5array14StreamOrDevice", "divmod"], [0, 1, 1, "_CPPv46divmodRK5arrayRK5array14StreamOrDevice", "divmod::a"], [0, 1, 1, "_CPPv46divmodRK5arrayRK5array14StreamOrDevice", "divmod::b"], [0, 1, 1, "_CPPv46divmodRK5arrayRK5array14StreamOrDevice", "divmod::s"], [0, 0, 1, "_CPPv45equalRK5arrayRK5array14StreamOrDevice", "equal"], [0, 1, 1, "_CPPv45equalRK5arrayRK5array14StreamOrDevice", "equal::a"], [0, 1, 1, "_CPPv45equalRK5arrayRK5array14StreamOrDevice", "equal::b"], [0, 1, 1, "_CPPv45equalRK5arrayRK5array14StreamOrDevice", "equal::s"], [0, 0, 1, "_CPPv43erfRK5array14StreamOrDevice", "erf"], [0, 1, 1, "_CPPv43erfRK5array14StreamOrDevice", "erf::a"], [0, 1, 1, "_CPPv43erfRK5array14StreamOrDevice", "erf::s"], [0, 0, 1, "_CPPv46erfinvRK5array14StreamOrDevice", "erfinv"], [0, 1, 1, "_CPPv46erfinvRK5array14StreamOrDevice", "erfinv::a"], [0, 1, 1, "_CPPv46erfinvRK5array14StreamOrDevice", "erfinv::s"], [0, 0, 1, "_CPPv43expRK5array14StreamOrDevice", "exp"], [0, 1, 1, "_CPPv43expRK5array14StreamOrDevice", "exp::a"], [0, 1, 1, "_CPPv43expRK5array14StreamOrDevice", "exp::s"], [0, 0, 1, "_CPPv411expand_dimsRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "expand_dims"], [0, 0, 1, "_CPPv411expand_dimsRK5arrayi14StreamOrDevice", "expand_dims"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "expand_dims::a"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayi14StreamOrDevice", "expand_dims::a"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "expand_dims::axes"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayi14StreamOrDevice", "expand_dims::axis"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "expand_dims::s"], [0, 1, 1, "_CPPv411expand_dimsRK5arrayi14StreamOrDevice", "expand_dims::s"], [0, 0, 1, "_CPPv45expm1RK5array14StreamOrDevice", "expm1"], [0, 1, 1, "_CPPv45expm1RK5array14StreamOrDevice", "expm1::a"], [0, 1, 1, "_CPPv45expm1RK5array14StreamOrDevice", "expm1::s"], [0, 0, 1, "_CPPv43eyei14StreamOrDevice", "eye"], [0, 0, 1, "_CPPv43eyei5Dtype14StreamOrDevice", "eye"], [0, 0, 1, "_CPPv43eyeii14StreamOrDevice", "eye"], [0, 0, 1, "_CPPv43eyeiii14StreamOrDevice", "eye"], [0, 0, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye"], [0, 1, 1, "_CPPv43eyei5Dtype14StreamOrDevice", "eye::dtype"], [0, 1, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye::dtype"], [0, 1, 1, "_CPPv43eyeiii14StreamOrDevice", "eye::k"], [0, 1, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye::k"], [0, 1, 1, "_CPPv43eyeii14StreamOrDevice", "eye::m"], [0, 1, 1, "_CPPv43eyeiii14StreamOrDevice", "eye::m"], [0, 1, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye::m"], [0, 1, 1, "_CPPv43eyei14StreamOrDevice", "eye::n"], [0, 1, 1, "_CPPv43eyei5Dtype14StreamOrDevice", "eye::n"], [0, 1, 1, "_CPPv43eyeii14StreamOrDevice", "eye::n"], [0, 1, 1, "_CPPv43eyeiii14StreamOrDevice", "eye::n"], [0, 1, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye::n"], [0, 1, 1, "_CPPv43eyei14StreamOrDevice", "eye::s"], [0, 1, 1, "_CPPv43eyei5Dtype14StreamOrDevice", "eye::s"], [0, 1, 1, "_CPPv43eyeii14StreamOrDevice", "eye::s"], [0, 1, 1, "_CPPv43eyeiii14StreamOrDevice", "eye::s"], [0, 1, 1, "_CPPv43eyeiii5Dtype14StreamOrDevice", "eye::s"], [0, 0, 1, "_CPPv47flattenRK5array14StreamOrDevice", "flatten"], [0, 0, 1, "_CPPv47flattenRK5arrayii14StreamOrDevice", "flatten"], [0, 1, 1, "_CPPv47flattenRK5array14StreamOrDevice", "flatten::a"], [0, 1, 1, "_CPPv47flattenRK5arrayii14StreamOrDevice", "flatten::a"], [0, 1, 1, "_CPPv47flattenRK5arrayii14StreamOrDevice", "flatten::end_axis"], [0, 1, 1, "_CPPv47flattenRK5array14StreamOrDevice", "flatten::s"], [0, 1, 1, "_CPPv47flattenRK5arrayii14StreamOrDevice", "flatten::s"], [0, 1, 1, "_CPPv47flattenRK5arrayii14StreamOrDevice", "flatten::start_axis"], [0, 0, 1, "_CPPv45floorRK5array14StreamOrDevice", "floor"], [0, 1, 1, "_CPPv45floorRK5array14StreamOrDevice", "floor::a"], [0, 1, 1, "_CPPv45floorRK5array14StreamOrDevice", "floor::s"], [0, 0, 1, "_CPPv412floor_divideRK5arrayRK5array14StreamOrDevice", "floor_divide"], [0, 1, 1, "_CPPv412floor_divideRK5arrayRK5array14StreamOrDevice", "floor_divide::a"], [0, 1, 1, "_CPPv412floor_divideRK5arrayRK5array14StreamOrDevice", "floor_divide::b"], [0, 1, 1, "_CPPv412floor_divideRK5arrayRK5array14StreamOrDevice", "floor_divide::s"], [0, 0, 1, "_CPPv44fullNSt6vectorIiEE5array14StreamOrDevice", "full"], [0, 0, 1, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", "full"], [0, 0, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", "full"], [0, 0, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full"], [0, 2, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", "full::T"], [0, 2, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full::T"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", "full::dtype"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full::dtype"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array14StreamOrDevice", "full::s"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", "full::s"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", "full::s"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full::s"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array14StreamOrDevice", "full::shape"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", "full::shape"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", "full::shape"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full::shape"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T14StreamOrDevice", "full::val"], [0, 1, 1, "_CPPv4I0E4full5arrayNSt6vectorIiEE1T5Dtype14StreamOrDevice", "full::val"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array14StreamOrDevice", "full::vals"], [0, 1, 1, "_CPPv44fullNSt6vectorIiEE5array5Dtype14StreamOrDevice", "full::vals"], [0, 0, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather"], [0, 0, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather"], [0, 1, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather::a"], [0, 1, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather::a"], [0, 1, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather::axes"], [0, 1, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather::axis"], [0, 1, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather::indices"], [0, 1, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather::indices"], [0, 1, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather::s"], [0, 1, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather::s"], [0, 1, 1, "_CPPv46gatherRK5arrayRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "gather::slice_sizes"], [0, 1, 1, "_CPPv46gatherRK5arrayRKNSt6vectorI5arrayEERKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "gather::slice_sizes"], [0, 0, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm"], [0, 1, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm::a"], [0, 1, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm::b"], [0, 1, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm::lhs_indices"], [0, 1, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm::rhs_indices"], [0, 1, 1, "_CPPv49gather_mm5array5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEE14StreamOrDevice", "gather_mm::s"], [0, 0, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::biases"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::bits"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::group_size"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::lhs_indices"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::rhs_indices"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::s"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::scales"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::transpose"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::w"], [0, 1, 1, "_CPPv410gather_qmmRK5arrayRK5arrayRK5arrayRK5arrayNSt8optionalI5arrayEENSt8optionalI5arrayEEbii14StreamOrDevice", "gather_qmm::x"], [0, 0, 1, "_CPPv47greaterRK5arrayRK5array14StreamOrDevice", "greater"], [0, 1, 1, "_CPPv47greaterRK5arrayRK5array14StreamOrDevice", "greater::a"], [0, 1, 1, "_CPPv47greaterRK5arrayRK5array14StreamOrDevice", "greater::b"], [0, 1, 1, "_CPPv47greaterRK5arrayRK5array14StreamOrDevice", "greater::s"], [0, 0, 1, "_CPPv413greater_equalRK5arrayRK5array14StreamOrDevice", "greater_equal"], [0, 1, 1, "_CPPv413greater_equalRK5arrayRK5array14StreamOrDevice", "greater_equal::a"], [0, 1, 1, "_CPPv413greater_equalRK5arrayRK5array14StreamOrDevice", "greater_equal::b"], [0, 1, 1, "_CPPv413greater_equalRK5arrayRK5array14StreamOrDevice", "greater_equal::s"], [0, 0, 1, "_CPPv418hadamard_transformRK5arrayNSt8optionalIfEE14StreamOrDevice", "hadamard_transform"], [0, 1, 1, "_CPPv418hadamard_transformRK5arrayNSt8optionalIfEE14StreamOrDevice", "hadamard_transform::a"], [0, 1, 1, "_CPPv418hadamard_transformRK5arrayNSt8optionalIfEE14StreamOrDevice", "hadamard_transform::s"], [0, 1, 1, "_CPPv418hadamard_transformRK5arrayNSt8optionalIfEE14StreamOrDevice", "hadamard_transform::scale"], [0, 0, 1, "_CPPv48identityi14StreamOrDevice", "identity"], [0, 0, 1, "_CPPv48identityi5Dtype14StreamOrDevice", "identity"], [0, 1, 1, "_CPPv48identityi5Dtype14StreamOrDevice", "identity::dtype"], [0, 1, 1, "_CPPv48identityi14StreamOrDevice", "identity::n"], [0, 1, 1, "_CPPv48identityi5Dtype14StreamOrDevice", "identity::n"], [0, 1, 1, "_CPPv48identityi14StreamOrDevice", "identity::s"], [0, 1, 1, "_CPPv48identityi5Dtype14StreamOrDevice", "identity::s"], [0, 0, 1, "_CPPv44imagRK5array14StreamOrDevice", "imag"], [0, 1, 1, "_CPPv44imagRK5array14StreamOrDevice", "imag::a"], [0, 1, 1, "_CPPv44imagRK5array14StreamOrDevice", "imag::s"], [0, 0, 1, "_CPPv45innerRK5arrayRK5array14StreamOrDevice", "inner"], [0, 1, 1, "_CPPv45innerRK5arrayRK5array14StreamOrDevice", "inner::a"], [0, 1, 1, "_CPPv45innerRK5arrayRK5array14StreamOrDevice", "inner::b"], [0, 1, 1, "_CPPv45innerRK5arrayRK5array14StreamOrDevice", "inner::s"], [0, 0, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::a"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::atol"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::b"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::equal_nan"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::rtol"], [0, 1, 1, "_CPPv47iscloseRK5arrayRK5arrayddb14StreamOrDevice", "isclose::s"], [0, 0, 1, "_CPPv48isfiniteRK5array14StreamOrDevice", "isfinite"], [0, 1, 1, "_CPPv48isfiniteRK5array14StreamOrDevice", "isfinite::a"], [0, 1, 1, "_CPPv48isfiniteRK5array14StreamOrDevice", "isfinite::s"], [0, 0, 1, "_CPPv45isinfRK5array14StreamOrDevice", "isinf"], [0, 1, 1, "_CPPv45isinfRK5array14StreamOrDevice", "isinf::a"], [0, 1, 1, "_CPPv45isinfRK5array14StreamOrDevice", "isinf::s"], [0, 0, 1, "_CPPv45isnanRK5array14StreamOrDevice", "isnan"], [0, 1, 1, "_CPPv45isnanRK5array14StreamOrDevice", "isnan::a"], [0, 1, 1, "_CPPv45isnanRK5array14StreamOrDevice", "isnan::s"], [0, 0, 1, "_CPPv48isneginfRK5array14StreamOrDevice", "isneginf"], [0, 1, 1, "_CPPv48isneginfRK5array14StreamOrDevice", "isneginf::a"], [0, 1, 1, "_CPPv48isneginfRK5array14StreamOrDevice", "isneginf::s"], [0, 0, 1, "_CPPv48isposinfRK5array14StreamOrDevice", "isposinf"], [0, 1, 1, "_CPPv48isposinfRK5array14StreamOrDevice", "isposinf::a"], [0, 1, 1, "_CPPv48isposinfRK5array14StreamOrDevice", "isposinf::s"], [0, 0, 1, "_CPPv410left_shiftRK5arrayRK5array14StreamOrDevice", "left_shift"], [0, 1, 1, "_CPPv410left_shiftRK5arrayRK5array14StreamOrDevice", "left_shift::a"], [0, 1, 1, "_CPPv410left_shiftRK5arrayRK5array14StreamOrDevice", "left_shift::b"], [0, 1, 1, "_CPPv410left_shiftRK5arrayRK5array14StreamOrDevice", "left_shift::s"], [0, 0, 1, "_CPPv44lessRK5arrayRK5array14StreamOrDevice", "less"], [0, 1, 1, "_CPPv44lessRK5arrayRK5array14StreamOrDevice", "less::a"], [0, 1, 1, "_CPPv44lessRK5arrayRK5array14StreamOrDevice", "less::b"], [0, 1, 1, "_CPPv44lessRK5arrayRK5array14StreamOrDevice", "less::s"], [0, 0, 1, "_CPPv410less_equalRK5arrayRK5array14StreamOrDevice", "less_equal"], [0, 1, 1, "_CPPv410less_equalRK5arrayRK5array14StreamOrDevice", "less_equal::a"], [0, 1, 1, "_CPPv410less_equalRK5arrayRK5array14StreamOrDevice", "less_equal::b"], [0, 1, 1, "_CPPv410less_equalRK5arrayRK5array14StreamOrDevice", "less_equal::s"], [0, 0, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace"], [0, 1, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace::dtype"], [0, 1, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace::num"], [0, 1, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace::s"], [0, 1, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace::start"], [0, 1, 1, "_CPPv48linspaceddi5Dtype14StreamOrDevice", "linspace::stop"], [0, 0, 1, "_CPPv43logRK5array14StreamOrDevice", "log"], [0, 0, 1, "_CPPv45log10RK5array14StreamOrDevice", "log10"], [0, 1, 1, "_CPPv45log10RK5array14StreamOrDevice", "log10::a"], [0, 1, 1, "_CPPv45log10RK5array14StreamOrDevice", "log10::s"], [0, 0, 1, "_CPPv45log1pRK5array14StreamOrDevice", "log1p"], [0, 1, 1, "_CPPv45log1pRK5array14StreamOrDevice", "log1p::a"], [0, 1, 1, "_CPPv45log1pRK5array14StreamOrDevice", "log1p::s"], [0, 0, 1, "_CPPv44log2RK5array14StreamOrDevice", "log2"], [0, 1, 1, "_CPPv44log2RK5array14StreamOrDevice", "log2::a"], [0, 1, 1, "_CPPv44log2RK5array14StreamOrDevice", "log2::s"], [0, 1, 1, "_CPPv43logRK5array14StreamOrDevice", "log::a"], [0, 1, 1, "_CPPv43logRK5array14StreamOrDevice", "log::s"], [0, 0, 1, "_CPPv49logaddexpRK5arrayRK5array14StreamOrDevice", "logaddexp"], [0, 1, 1, "_CPPv49logaddexpRK5arrayRK5array14StreamOrDevice", "logaddexp::a"], [0, 1, 1, "_CPPv49logaddexpRK5arrayRK5array14StreamOrDevice", "logaddexp::b"], [0, 1, 1, "_CPPv49logaddexpRK5arrayRK5array14StreamOrDevice", "logaddexp::s"], [0, 0, 1, "_CPPv411logical_andRK5arrayRK5array14StreamOrDevice", "logical_and"], [0, 1, 1, "_CPPv411logical_andRK5arrayRK5array14StreamOrDevice", "logical_and::a"], [0, 1, 1, "_CPPv411logical_andRK5arrayRK5array14StreamOrDevice", "logical_and::b"], [0, 1, 1, "_CPPv411logical_andRK5arrayRK5array14StreamOrDevice", "logical_and::s"], [0, 0, 1, "_CPPv411logical_notRK5array14StreamOrDevice", "logical_not"], [0, 1, 1, "_CPPv411logical_notRK5array14StreamOrDevice", "logical_not::a"], [0, 1, 1, "_CPPv411logical_notRK5array14StreamOrDevice", "logical_not::s"], [0, 0, 1, "_CPPv410logical_orRK5arrayRK5array14StreamOrDevice", "logical_or"], [0, 1, 1, "_CPPv410logical_orRK5arrayRK5array14StreamOrDevice", "logical_or::a"], [0, 1, 1, "_CPPv410logical_orRK5arrayRK5array14StreamOrDevice", "logical_or::b"], [0, 1, 1, "_CPPv410logical_orRK5arrayRK5array14StreamOrDevice", "logical_or::s"], [0, 0, 1, "_CPPv49logsumexpRK5array14StreamOrDevice", "logsumexp"], [0, 0, 1, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "logsumexp"], [0, 0, 1, "_CPPv49logsumexpRK5arrayb14StreamOrDevice", "logsumexp"], [0, 0, 1, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", "logsumexp"], [0, 1, 1, "_CPPv49logsumexpRK5array14StreamOrDevice", "logsumexp::a"], [0, 1, 1, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "logsumexp::a"], [0, 1, 1, "_CPPv49logsumexpRK5arrayb14StreamOrDevice", "logsumexp::a"], [0, 1, 1, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", "logsumexp::a"], [0, 1, 1, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "logsumexp::axes"], [0, 1, 1, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", "logsumexp::axis"], [0, 1, 1, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "logsumexp::keepdims"], [0, 1, 1, "_CPPv49logsumexpRK5arrayb14StreamOrDevice", "logsumexp::keepdims"], [0, 1, 1, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", "logsumexp::keepdims"], [0, 1, 1, "_CPPv49logsumexpRK5array14StreamOrDevice", "logsumexp::s"], [0, 1, 1, "_CPPv49logsumexpRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "logsumexp::s"], [0, 1, 1, "_CPPv49logsumexpRK5arrayb14StreamOrDevice", "logsumexp::s"], [0, 1, 1, "_CPPv49logsumexpRK5arrayib14StreamOrDevice", "logsumexp::s"], [0, 0, 1, "_CPPv46matmulRK5arrayRK5array14StreamOrDevice", "matmul"], [0, 1, 1, "_CPPv46matmulRK5arrayRK5array14StreamOrDevice", "matmul::a"], [0, 1, 1, "_CPPv46matmulRK5arrayRK5array14StreamOrDevice", "matmul::b"], [0, 1, 1, "_CPPv46matmulRK5arrayRK5array14StreamOrDevice", "matmul::s"], [0, 0, 1, "_CPPv43maxRK5array14StreamOrDevice", "max"], [0, 0, 1, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "max"], [0, 0, 1, "_CPPv43maxRK5arrayb14StreamOrDevice", "max"], [0, 0, 1, "_CPPv43maxRK5arrayib14StreamOrDevice", "max"], [0, 1, 1, "_CPPv43maxRK5array14StreamOrDevice", "max::a"], [0, 1, 1, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "max::a"], [0, 1, 1, "_CPPv43maxRK5arrayb14StreamOrDevice", "max::a"], [0, 1, 1, "_CPPv43maxRK5arrayib14StreamOrDevice", "max::a"], [0, 1, 1, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "max::axes"], [0, 1, 1, "_CPPv43maxRK5arrayib14StreamOrDevice", "max::axis"], [0, 1, 1, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "max::keepdims"], [0, 1, 1, "_CPPv43maxRK5arrayb14StreamOrDevice", "max::keepdims"], [0, 1, 1, "_CPPv43maxRK5arrayib14StreamOrDevice", "max::keepdims"], [0, 1, 1, "_CPPv43maxRK5array14StreamOrDevice", "max::s"], [0, 1, 1, "_CPPv43maxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "max::s"], [0, 1, 1, "_CPPv43maxRK5arrayb14StreamOrDevice", "max::s"], [0, 1, 1, "_CPPv43maxRK5arrayib14StreamOrDevice", "max::s"], [0, 0, 1, "_CPPv47maximumRK5arrayRK5array14StreamOrDevice", "maximum"], [0, 1, 1, "_CPPv47maximumRK5arrayRK5array14StreamOrDevice", "maximum::a"], [0, 1, 1, "_CPPv47maximumRK5arrayRK5array14StreamOrDevice", "maximum::b"], [0, 1, 1, "_CPPv47maximumRK5arrayRK5array14StreamOrDevice", "maximum::s"], [0, 0, 1, "_CPPv44meanRK5array14StreamOrDevice", "mean"], [0, 0, 1, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "mean"], [0, 0, 1, "_CPPv44meanRK5arrayb14StreamOrDevice", "mean"], [0, 0, 1, "_CPPv44meanRK5arrayib14StreamOrDevice", "mean"], [0, 1, 1, "_CPPv44meanRK5array14StreamOrDevice", "mean::a"], [0, 1, 1, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "mean::a"], [0, 1, 1, "_CPPv44meanRK5arrayb14StreamOrDevice", "mean::a"], [0, 1, 1, "_CPPv44meanRK5arrayib14StreamOrDevice", "mean::a"], [0, 1, 1, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "mean::axes"], [0, 1, 1, "_CPPv44meanRK5arrayib14StreamOrDevice", "mean::axis"], [0, 1, 1, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "mean::keepdims"], [0, 1, 1, "_CPPv44meanRK5arrayb14StreamOrDevice", "mean::keepdims"], [0, 1, 1, "_CPPv44meanRK5arrayib14StreamOrDevice", "mean::keepdims"], [0, 1, 1, "_CPPv44meanRK5array14StreamOrDevice", "mean::s"], [0, 1, 1, "_CPPv44meanRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "mean::s"], [0, 1, 1, "_CPPv44meanRK5arrayb14StreamOrDevice", "mean::s"], [0, 1, 1, "_CPPv44meanRK5arrayib14StreamOrDevice", "mean::s"], [0, 0, 1, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", "meshgrid"], [0, 1, 1, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", "meshgrid::arrays"], [0, 1, 1, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", "meshgrid::indexing"], [0, 1, 1, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", "meshgrid::s"], [0, 1, 1, "_CPPv48meshgridRKNSt6vectorI5arrayEEbNSt6stringE14StreamOrDevice", "meshgrid::sparse"], [0, 0, 1, "_CPPv43minRK5array14StreamOrDevice", "min"], [0, 0, 1, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "min"], [0, 0, 1, "_CPPv43minRK5arrayb14StreamOrDevice", "min"], [0, 0, 1, "_CPPv43minRK5arrayib14StreamOrDevice", "min"], [0, 1, 1, "_CPPv43minRK5array14StreamOrDevice", "min::a"], [0, 1, 1, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "min::a"], [0, 1, 1, "_CPPv43minRK5arrayb14StreamOrDevice", "min::a"], [0, 1, 1, "_CPPv43minRK5arrayib14StreamOrDevice", "min::a"], [0, 1, 1, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "min::axes"], [0, 1, 1, "_CPPv43minRK5arrayib14StreamOrDevice", "min::axis"], [0, 1, 1, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "min::keepdims"], [0, 1, 1, "_CPPv43minRK5arrayb14StreamOrDevice", "min::keepdims"], [0, 1, 1, "_CPPv43minRK5arrayib14StreamOrDevice", "min::keepdims"], [0, 1, 1, "_CPPv43minRK5array14StreamOrDevice", "min::s"], [0, 1, 1, "_CPPv43minRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "min::s"], [0, 1, 1, "_CPPv43minRK5arrayb14StreamOrDevice", "min::s"], [0, 1, 1, "_CPPv43minRK5arrayib14StreamOrDevice", "min::s"], [0, 0, 1, "_CPPv47minimumRK5arrayRK5array14StreamOrDevice", "minimum"], [0, 1, 1, "_CPPv47minimumRK5arrayRK5array14StreamOrDevice", "minimum::a"], [0, 1, 1, "_CPPv47minimumRK5arrayRK5array14StreamOrDevice", "minimum::b"], [0, 1, 1, "_CPPv47minimumRK5arrayRK5array14StreamOrDevice", "minimum::s"], [0, 0, 1, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", "moveaxis"], [0, 1, 1, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", "moveaxis::a"], [0, 1, 1, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", "moveaxis::destination"], [0, 1, 1, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", "moveaxis::s"], [0, 1, 1, "_CPPv48moveaxisRK5arrayii14StreamOrDevice", "moveaxis::source"], [0, 0, 1, "_CPPv48multiplyRK5arrayRK5array14StreamOrDevice", "multiply"], [0, 1, 1, "_CPPv48multiplyRK5arrayRK5array14StreamOrDevice", "multiply::a"], [0, 1, 1, "_CPPv48multiplyRK5arrayRK5array14StreamOrDevice", "multiply::b"], [0, 1, 1, "_CPPv48multiplyRK5arrayRK5array14StreamOrDevice", "multiply::s"], [0, 0, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num"], [0, 1, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num::a"], [0, 1, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num::nan"], [0, 1, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num::neginf"], [0, 1, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num::posinf"], [0, 1, 1, "_CPPv410nan_to_numRK5arrayfKNSt8optionalIfEEKNSt8optionalIfEE14StreamOrDevice", "nan_to_num::s"], [0, 0, 1, "_CPPv48negativeRK5array14StreamOrDevice", "negative"], [0, 1, 1, "_CPPv48negativeRK5array14StreamOrDevice", "negative::a"], [0, 1, 1, "_CPPv48negativeRK5array14StreamOrDevice", "negative::s"], [0, 0, 1, "_CPPv49not_equalRK5arrayRK5array14StreamOrDevice", "not_equal"], [0, 1, 1, "_CPPv49not_equalRK5arrayRK5array14StreamOrDevice", "not_equal::a"], [0, 1, 1, "_CPPv49not_equalRK5arrayRK5array14StreamOrDevice", "not_equal::b"], [0, 1, 1, "_CPPv49not_equalRK5arrayRK5array14StreamOrDevice", "not_equal::s"], [0, 0, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements"], [0, 1, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements::a"], [0, 1, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements::axes"], [0, 1, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements::dtype"], [0, 1, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements::inverted"], [0, 1, 1, "_CPPv418number_of_elementsRK5arrayNSt6vectorIiEEb5Dtype14StreamOrDevice", "number_of_elements::s"], [0, 0, 1, "_CPPv44onesRKNSt6vectorIiEE14StreamOrDevice", "ones"], [0, 0, 1, "_CPPv44onesRKNSt6vectorIiEE5Dtype14StreamOrDevice", "ones"], [0, 1, 1, "_CPPv44onesRKNSt6vectorIiEE5Dtype14StreamOrDevice", "ones::dtype"], [0, 1, 1, "_CPPv44onesRKNSt6vectorIiEE14StreamOrDevice", "ones::s"], [0, 1, 1, "_CPPv44onesRKNSt6vectorIiEE5Dtype14StreamOrDevice", "ones::s"], [0, 1, 1, "_CPPv44onesRKNSt6vectorIiEE14StreamOrDevice", "ones::shape"], [0, 1, 1, "_CPPv44onesRKNSt6vectorIiEE5Dtype14StreamOrDevice", "ones::shape"], [0, 0, 1, "_CPPv49ones_likeRK5array14StreamOrDevice", "ones_like"], [0, 1, 1, "_CPPv49ones_likeRK5array14StreamOrDevice", "ones_like::a"], [0, 1, 1, "_CPPv49ones_likeRK5array14StreamOrDevice", "ones_like::s"], [0, 0, 1, "_CPPv4I0Ene5array1TRK5array", "operator!="], [0, 0, 1, "_CPPv4I0Ene5arrayRK5array1T", "operator!="], [0, 0, 1, "_CPPv4neRK5arrayRK5array", "operator!="], [0, 2, 1, "_CPPv4I0Ene5array1TRK5array", "operator!=::T"], [0, 2, 1, "_CPPv4I0Ene5arrayRK5array1T", "operator!=::T"], [0, 1, 1, "_CPPv4I0Ene5array1TRK5array", "operator!=::a"], [0, 1, 1, "_CPPv4I0Ene5arrayRK5array1T", "operator!=::a"], [0, 1, 1, "_CPPv4neRK5arrayRK5array", "operator!=::a"], [0, 1, 1, "_CPPv4I0Ene5array1TRK5array", "operator!=::b"], [0, 1, 1, "_CPPv4I0Ene5arrayRK5array1T", "operator!=::b"], [0, 1, 1, "_CPPv4neRK5arrayRK5array", "operator!=::b"], [0, 0, 1, "_CPPv4I0Erm5array1TRK5array", "operator%"], [0, 0, 1, "_CPPv4I0Erm5arrayRK5array1T", "operator%"], [0, 0, 1, "_CPPv4rmRK5arrayRK5array", "operator%"], [0, 2, 1, "_CPPv4I0Erm5array1TRK5array", "operator%::T"], [0, 2, 1, "_CPPv4I0Erm5arrayRK5array1T", "operator%::T"], [0, 1, 1, "_CPPv4I0Erm5array1TRK5array", "operator%::a"], [0, 1, 1, "_CPPv4I0Erm5arrayRK5array1T", "operator%::a"], [0, 1, 1, "_CPPv4rmRK5arrayRK5array", "operator%::a"], [0, 1, 1, "_CPPv4I0Erm5array1TRK5array", "operator%::b"], [0, 1, 1, "_CPPv4I0Erm5arrayRK5array1T", "operator%::b"], [0, 1, 1, "_CPPv4rmRK5arrayRK5array", "operator%::b"], [0, 0, 1, "_CPPv4anRK5arrayRK5array", "operator&"], [0, 0, 1, "_CPPv4aaRK5arrayRK5array", "operator&&"], [0, 1, 1, "_CPPv4aaRK5arrayRK5array", "operator&&::a"], [0, 1, 1, "_CPPv4aaRK5arrayRK5array", "operator&&::b"], [0, 1, 1, "_CPPv4anRK5arrayRK5array", "operator&::a"], [0, 1, 1, "_CPPv4anRK5arrayRK5array", "operator&::b"], [0, 0, 1, "_CPPv4I0Eml5array1TRK5array", "operator*"], [0, 0, 1, "_CPPv4I0Eml5arrayRK5array1T", "operator*"], [0, 0, 1, "_CPPv4mlRK5arrayRK5array", "operator*"], [0, 2, 1, "_CPPv4I0Eml5array1TRK5array", "operator*::T"], [0, 2, 1, "_CPPv4I0Eml5arrayRK5array1T", "operator*::T"], [0, 1, 1, "_CPPv4I0Eml5array1TRK5array", "operator*::a"], [0, 1, 1, "_CPPv4I0Eml5arrayRK5array1T", "operator*::a"], [0, 1, 1, "_CPPv4mlRK5arrayRK5array", "operator*::a"], [0, 1, 1, "_CPPv4I0Eml5array1TRK5array", "operator*::b"], [0, 1, 1, "_CPPv4I0Eml5arrayRK5array1T", "operator*::b"], [0, 1, 1, "_CPPv4mlRK5arrayRK5array", "operator*::b"], [0, 0, 1, "_CPPv4I0Epl5array1TRK5array", "operator+"], [0, 0, 1, "_CPPv4I0Epl5arrayRK5array1T", "operator+"], [0, 0, 1, "_CPPv4plRK5arrayRK5array", "operator+"], [0, 2, 1, "_CPPv4I0Epl5array1TRK5array", "operator+::T"], [0, 2, 1, "_CPPv4I0Epl5arrayRK5array1T", "operator+::T"], [0, 1, 1, "_CPPv4I0Epl5array1TRK5array", "operator+::a"], [0, 1, 1, "_CPPv4I0Epl5arrayRK5array1T", "operator+::a"], [0, 1, 1, "_CPPv4plRK5arrayRK5array", "operator+::a"], [0, 1, 1, "_CPPv4I0Epl5array1TRK5array", "operator+::b"], [0, 1, 1, "_CPPv4I0Epl5arrayRK5array1T", "operator+::b"], [0, 1, 1, "_CPPv4plRK5arrayRK5array", "operator+::b"], [0, 0, 1, "_CPPv4I0Emi5array1TRK5array", "operator-"], [0, 0, 1, "_CPPv4I0Emi5arrayRK5array1T", "operator-"], [0, 0, 1, "_CPPv4miRK5array", "operator-"], [0, 0, 1, "_CPPv4miRK5arrayRK5array", "operator-"], [0, 2, 1, "_CPPv4I0Emi5array1TRK5array", "operator-::T"], [0, 2, 1, "_CPPv4I0Emi5arrayRK5array1T", "operator-::T"], [0, 1, 1, "_CPPv4I0Emi5array1TRK5array", "operator-::a"], [0, 1, 1, "_CPPv4I0Emi5arrayRK5array1T", "operator-::a"], [0, 1, 1, "_CPPv4miRK5array", "operator-::a"], [0, 1, 1, "_CPPv4miRK5arrayRK5array", "operator-::a"], [0, 1, 1, "_CPPv4I0Emi5array1TRK5array", "operator-::b"], [0, 1, 1, "_CPPv4I0Emi5arrayRK5array1T", "operator-::b"], [0, 1, 1, "_CPPv4miRK5arrayRK5array", "operator-::b"], [0, 0, 1, "_CPPv4dvRK5arrayRK5array", "operator/"], [0, 0, 1, "_CPPv4dvRK5arrayd", "operator/"], [0, 0, 1, "_CPPv4dvdRK5array", "operator/"], [0, 1, 1, "_CPPv4dvRK5arrayRK5array", "operator/::a"], [0, 1, 1, "_CPPv4dvRK5arrayd", "operator/::a"], [0, 1, 1, "_CPPv4dvdRK5array", "operator/::a"], [0, 1, 1, "_CPPv4dvRK5arrayRK5array", "operator/::b"], [0, 1, 1, "_CPPv4dvRK5arrayd", "operator/::b"], [0, 1, 1, "_CPPv4dvdRK5array", "operator/::b"], [0, 0, 1, "_CPPv4I0Elt5array1TRK5array", "operator<"], [0, 0, 1, "_CPPv4I0Elt5arrayRK5array1T", "operator<"], [0, 0, 1, "_CPPv4ltRK5arrayRK5array", "operator<"], [0, 2, 1, "_CPPv4I0Elt5array1TRK5array", "operator<::T"], [0, 2, 1, "_CPPv4I0Elt5arrayRK5array1T", "operator<::T"], [0, 1, 1, "_CPPv4I0Elt5array1TRK5array", "operator<::a"], [0, 1, 1, "_CPPv4I0Elt5arrayRK5array1T", "operator<::a"], [0, 1, 1, "_CPPv4ltRK5arrayRK5array", "operator<::a"], [0, 1, 1, "_CPPv4I0Elt5array1TRK5array", "operator<::b"], [0, 1, 1, "_CPPv4I0Elt5arrayRK5array1T", "operator<::b"], [0, 1, 1, "_CPPv4ltRK5arrayRK5array", "operator<::b"], [0, 0, 1, "_CPPv4lsRK5arrayRK5array", "operator<<"], [0, 1, 1, "_CPPv4lsRK5arrayRK5array", "operator<<::a"], [0, 1, 1, "_CPPv4lsRK5arrayRK5array", "operator<<::b"], [0, 0, 1, "_CPPv4I0Ele5array1TRK5array", "operator<="], [0, 0, 1, "_CPPv4I0Ele5arrayRK5array1T", "operator<="], [0, 0, 1, "_CPPv4leRK5arrayRK5array", "operator<="], [0, 2, 1, "_CPPv4I0Ele5array1TRK5array", "operator<=::T"], [0, 2, 1, "_CPPv4I0Ele5arrayRK5array1T", "operator<=::T"], [0, 1, 1, "_CPPv4I0Ele5array1TRK5array", "operator<=::a"], [0, 1, 1, "_CPPv4I0Ele5arrayRK5array1T", "operator<=::a"], [0, 1, 1, "_CPPv4leRK5arrayRK5array", "operator<=::a"], [0, 1, 1, "_CPPv4I0Ele5array1TRK5array", "operator<=::b"], [0, 1, 1, "_CPPv4I0Ele5arrayRK5array1T", "operator<=::b"], [0, 1, 1, "_CPPv4leRK5arrayRK5array", "operator<=::b"], [0, 0, 1, "_CPPv4I0Eeq5array1TRK5array", "operator=="], [0, 0, 1, "_CPPv4I0Eeq5arrayRK5array1T", "operator=="], [0, 0, 1, "_CPPv4eqRK5arrayRK5array", "operator=="], [0, 2, 1, "_CPPv4I0Eeq5array1TRK5array", "operator==::T"], [0, 2, 1, "_CPPv4I0Eeq5arrayRK5array1T", "operator==::T"], [0, 1, 1, "_CPPv4I0Eeq5array1TRK5array", "operator==::a"], [0, 1, 1, "_CPPv4I0Eeq5arrayRK5array1T", "operator==::a"], [0, 1, 1, "_CPPv4eqRK5arrayRK5array", "operator==::a"], [0, 1, 1, "_CPPv4I0Eeq5array1TRK5array", "operator==::b"], [0, 1, 1, "_CPPv4I0Eeq5arrayRK5array1T", "operator==::b"], [0, 1, 1, "_CPPv4eqRK5arrayRK5array", "operator==::b"], [0, 0, 1, "_CPPv4I0Egt5array1TRK5array", "operator>"], [0, 0, 1, "_CPPv4I0Egt5arrayRK5array1T", "operator>"], [0, 0, 1, "_CPPv4gtRK5arrayRK5array", "operator>"], [0, 2, 1, "_CPPv4I0Egt5array1TRK5array", "operator>::T"], [0, 2, 1, "_CPPv4I0Egt5arrayRK5array1T", "operator>::T"], [0, 1, 1, "_CPPv4I0Egt5array1TRK5array", "operator>::a"], [0, 1, 1, "_CPPv4I0Egt5arrayRK5array1T", "operator>::a"], [0, 1, 1, "_CPPv4gtRK5arrayRK5array", "operator>::a"], [0, 1, 1, "_CPPv4I0Egt5array1TRK5array", "operator>::b"], [0, 1, 1, "_CPPv4I0Egt5arrayRK5array1T", "operator>::b"], [0, 1, 1, "_CPPv4gtRK5arrayRK5array", "operator>::b"], [0, 0, 1, "_CPPv4I0Ege5array1TRK5array", "operator>="], [0, 0, 1, "_CPPv4I0Ege5arrayRK5array1T", "operator>="], [0, 0, 1, "_CPPv4geRK5arrayRK5array", "operator>="], [0, 2, 1, "_CPPv4I0Ege5array1TRK5array", "operator>=::T"], [0, 2, 1, "_CPPv4I0Ege5arrayRK5array1T", "operator>=::T"], [0, 1, 1, "_CPPv4I0Ege5array1TRK5array", "operator>=::a"], [0, 1, 1, "_CPPv4I0Ege5arrayRK5array1T", "operator>=::a"], [0, 1, 1, "_CPPv4geRK5arrayRK5array", "operator>=::a"], [0, 1, 1, "_CPPv4I0Ege5array1TRK5array", "operator>=::b"], [0, 1, 1, "_CPPv4I0Ege5arrayRK5array1T", "operator>=::b"], [0, 1, 1, "_CPPv4geRK5arrayRK5array", "operator>=::b"], [0, 0, 1, "_CPPv4rsRK5arrayRK5array", "operator>>"], [0, 1, 1, "_CPPv4rsRK5arrayRK5array", "operator>>::a"], [0, 1, 1, "_CPPv4rsRK5arrayRK5array", "operator>>::b"], [0, 0, 1, "_CPPv4eoRK5arrayRK5array", "operator^"], [0, 1, 1, "_CPPv4eoRK5arrayRK5array", "operator^::a"], [0, 1, 1, "_CPPv4eoRK5arrayRK5array", "operator^::b"], [0, 0, 1, "_CPPv4orRK5arrayRK5array", "operator|"], [0, 1, 1, "_CPPv4orRK5arrayRK5array", "operator|::a"], [0, 1, 1, "_CPPv4orRK5arrayRK5array", "operator|::b"], [0, 0, 1, "_CPPv4ooRK5arrayRK5array", "operator||"], [0, 1, 1, "_CPPv4ooRK5arrayRK5array", "operator||::a"], [0, 1, 1, "_CPPv4ooRK5arrayRK5array", "operator||::b"], [0, 0, 1, "_CPPv45outerRK5arrayRK5array14StreamOrDevice", "outer"], [0, 1, 1, "_CPPv45outerRK5arrayRK5array14StreamOrDevice", "outer::a"], [0, 1, 1, "_CPPv45outerRK5arrayRK5array14StreamOrDevice", "outer::b"], [0, 1, 1, "_CPPv45outerRK5arrayRK5array14StreamOrDevice", "outer::s"], [0, 0, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad"], [0, 0, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad"], [0, 0, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad"], [0, 0, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::a"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad::a"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::a"], [0, 1, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad::a"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::axes"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::high_pad_size"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::low_pad_size"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::mode"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad::mode"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::mode"], [0, 1, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad::mode"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_value"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_value"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_value"], [0, 1, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_value"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_width"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_width"], [0, 1, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad::pad_width"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt4pairIiiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::s"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorINSt4pairIiiEEEERK5arrayKNSt6stringE14StreamOrDevice", "pad::s"], [0, 1, 1, "_CPPv43padRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEERKNSt6vectorIiEERK5arrayKNSt6stringE14StreamOrDevice", "pad::s"], [0, 1, 1, "_CPPv43padRK5arrayiRK5arrayKNSt6stringE14StreamOrDevice", "pad::s"], [0, 0, 1, "_CPPv49partitionRK5arrayi14StreamOrDevice", "partition"], [0, 0, 1, "_CPPv49partitionRK5arrayii14StreamOrDevice", "partition"], [0, 1, 1, "_CPPv49partitionRK5arrayi14StreamOrDevice", "partition::a"], [0, 1, 1, "_CPPv49partitionRK5arrayii14StreamOrDevice", "partition::a"], [0, 1, 1, "_CPPv49partitionRK5arrayii14StreamOrDevice", "partition::axis"], [0, 1, 1, "_CPPv49partitionRK5arrayi14StreamOrDevice", "partition::kth"], [0, 1, 1, "_CPPv49partitionRK5arrayii14StreamOrDevice", "partition::kth"], [0, 1, 1, "_CPPv49partitionRK5arrayi14StreamOrDevice", "partition::s"], [0, 1, 1, "_CPPv49partitionRK5arrayii14StreamOrDevice", "partition::s"], [0, 0, 1, "_CPPv45powerRK5arrayRK5array14StreamOrDevice", "power"], [0, 1, 1, "_CPPv45powerRK5arrayRK5array14StreamOrDevice", "power::a"], [0, 1, 1, "_CPPv45powerRK5arrayRK5array14StreamOrDevice", "power::b"], [0, 1, 1, "_CPPv45powerRK5arrayRK5array14StreamOrDevice", "power::s"], [0, 0, 1, "_CPPv44prodRK5array14StreamOrDevice", "prod"], [0, 0, 1, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "prod"], [0, 0, 1, "_CPPv44prodRK5arrayb14StreamOrDevice", "prod"], [0, 0, 1, "_CPPv44prodRK5arrayib14StreamOrDevice", "prod"], [0, 1, 1, "_CPPv44prodRK5array14StreamOrDevice", "prod::a"], [0, 1, 1, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "prod::a"], [0, 1, 1, "_CPPv44prodRK5arrayb14StreamOrDevice", "prod::a"], [0, 1, 1, "_CPPv44prodRK5arrayib14StreamOrDevice", "prod::a"], [0, 1, 1, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "prod::axes"], [0, 1, 1, "_CPPv44prodRK5arrayib14StreamOrDevice", "prod::axis"], [0, 1, 1, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "prod::keepdims"], [0, 1, 1, "_CPPv44prodRK5arrayb14StreamOrDevice", "prod::keepdims"], [0, 1, 1, "_CPPv44prodRK5arrayib14StreamOrDevice", "prod::keepdims"], [0, 1, 1, "_CPPv44prodRK5array14StreamOrDevice", "prod::s"], [0, 1, 1, "_CPPv44prodRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "prod::s"], [0, 1, 1, "_CPPv44prodRK5arrayb14StreamOrDevice", "prod::s"], [0, 1, 1, "_CPPv44prodRK5arrayib14StreamOrDevice", "prod::s"], [0, 0, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis"], [0, 1, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis::a"], [0, 1, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis::axis"], [0, 1, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis::indices"], [0, 1, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis::s"], [0, 1, 1, "_CPPv414put_along_axisRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "put_along_axis::values"], [0, 0, 1, "_CPPv48quantizeRK5arrayii14StreamOrDevice", "quantize"], [0, 1, 1, "_CPPv48quantizeRK5arrayii14StreamOrDevice", "quantize::bits"], [0, 1, 1, "_CPPv48quantizeRK5arrayii14StreamOrDevice", "quantize::group_size"], [0, 1, 1, "_CPPv48quantizeRK5arrayii14StreamOrDevice", "quantize::s"], [0, 1, 1, "_CPPv48quantizeRK5arrayii14StreamOrDevice", "quantize::w"], [0, 0, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::biases"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::bits"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::group_size"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::s"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::scales"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::transpose"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::w"], [0, 1, 1, "_CPPv416quantized_matmul5array5array5array5arraybii14StreamOrDevice", "quantized_matmul::x"], [0, 0, 1, "_CPPv47radiansRK5array14StreamOrDevice", "radians"], [0, 1, 1, "_CPPv47radiansRK5array14StreamOrDevice", "radians::a"], [0, 1, 1, "_CPPv47radiansRK5array14StreamOrDevice", "radians::s"], [0, 0, 1, "_CPPv44realRK5array14StreamOrDevice", "real"], [0, 1, 1, "_CPPv44realRK5array14StreamOrDevice", "real::a"], [0, 1, 1, "_CPPv44realRK5array14StreamOrDevice", "real::s"], [0, 0, 1, "_CPPv410reciprocalRK5array14StreamOrDevice", "reciprocal"], [0, 1, 1, "_CPPv410reciprocalRK5array14StreamOrDevice", "reciprocal::a"], [0, 1, 1, "_CPPv410reciprocalRK5array14StreamOrDevice", "reciprocal::s"], [0, 0, 1, "_CPPv49remainderRK5arrayRK5array14StreamOrDevice", "remainder"], [0, 1, 1, "_CPPv49remainderRK5arrayRK5array14StreamOrDevice", "remainder::a"], [0, 1, 1, "_CPPv49remainderRK5arrayRK5array14StreamOrDevice", "remainder::b"], [0, 1, 1, "_CPPv49remainderRK5arrayRK5array14StreamOrDevice", "remainder::s"], [0, 0, 1, "_CPPv46repeatRK5arrayi14StreamOrDevice", "repeat"], [0, 0, 1, "_CPPv46repeatRK5arrayii14StreamOrDevice", "repeat"], [0, 1, 1, "_CPPv46repeatRK5arrayi14StreamOrDevice", "repeat::arr"], [0, 1, 1, "_CPPv46repeatRK5arrayii14StreamOrDevice", "repeat::arr"], [0, 1, 1, "_CPPv46repeatRK5arrayii14StreamOrDevice", "repeat::axis"], [0, 1, 1, "_CPPv46repeatRK5arrayi14StreamOrDevice", "repeat::repeats"], [0, 1, 1, "_CPPv46repeatRK5arrayii14StreamOrDevice", "repeat::repeats"], [0, 1, 1, "_CPPv46repeatRK5arrayi14StreamOrDevice", "repeat::s"], [0, 1, 1, "_CPPv46repeatRK5arrayii14StreamOrDevice", "repeat::s"], [0, 0, 1, "_CPPv47reshapeRK5arrayNSt6vectorIiEE14StreamOrDevice", "reshape"], [0, 1, 1, "_CPPv47reshapeRK5arrayNSt6vectorIiEE14StreamOrDevice", "reshape::a"], [0, 1, 1, "_CPPv47reshapeRK5arrayNSt6vectorIiEE14StreamOrDevice", "reshape::s"], [0, 1, 1, "_CPPv47reshapeRK5arrayNSt6vectorIiEE14StreamOrDevice", "reshape::shape"], [0, 0, 1, "_CPPv411right_shiftRK5arrayRK5array14StreamOrDevice", "right_shift"], [0, 1, 1, "_CPPv411right_shiftRK5arrayRK5array14StreamOrDevice", "right_shift::a"], [0, 1, 1, "_CPPv411right_shiftRK5arrayRK5array14StreamOrDevice", "right_shift::b"], [0, 1, 1, "_CPPv411right_shiftRK5arrayRK5array14StreamOrDevice", "right_shift::s"], [0, 0, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "roll"], [0, 0, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "roll"], [0, 0, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "roll"], [0, 0, 1, "_CPPv44rollRK5arrayi14StreamOrDevice", "roll"], [0, 0, 1, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "roll"], [0, 0, 1, "_CPPv44rollRK5arrayii14StreamOrDevice", "roll"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayi14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayii14StreamOrDevice", "roll::a"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "roll::axes"], [0, 1, 1, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "roll::axes"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "roll::axis"], [0, 1, 1, "_CPPv44rollRK5arrayii14StreamOrDevice", "roll::axis"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayi14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayii14StreamOrDevice", "roll::s"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "roll::shift"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "roll::shift"], [0, 1, 1, "_CPPv44rollRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "roll::shift"], [0, 1, 1, "_CPPv44rollRK5arrayi14StreamOrDevice", "roll::shift"], [0, 1, 1, "_CPPv44rollRK5arrayiRKNSt6vectorIiEE14StreamOrDevice", "roll::shift"], [0, 1, 1, "_CPPv44rollRK5arrayii14StreamOrDevice", "roll::shift"], [0, 0, 1, "_CPPv45roundRK5array14StreamOrDevice", "round"], [0, 0, 1, "_CPPv45roundRK5arrayi14StreamOrDevice", "round"], [0, 1, 1, "_CPPv45roundRK5array14StreamOrDevice", "round::a"], [0, 1, 1, "_CPPv45roundRK5arrayi14StreamOrDevice", "round::a"], [0, 1, 1, "_CPPv45roundRK5arrayi14StreamOrDevice", "round::decimals"], [0, 1, 1, "_CPPv45roundRK5array14StreamOrDevice", "round::s"], [0, 1, 1, "_CPPv45roundRK5arrayi14StreamOrDevice", "round::s"], [0, 0, 1, "_CPPv45rsqrtRK5array14StreamOrDevice", "rsqrt"], [0, 1, 1, "_CPPv45rsqrtRK5array14StreamOrDevice", "rsqrt::a"], [0, 1, 1, "_CPPv45rsqrtRK5array14StreamOrDevice", "rsqrt::s"], [0, 0, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter"], [0, 0, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter"], [0, 1, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter::a"], [0, 1, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter::a"], [0, 1, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter::axes"], [0, 1, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter::axis"], [0, 1, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter::indices"], [0, 1, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter::indices"], [0, 1, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter::s"], [0, 1, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter::s"], [0, 1, 1, "_CPPv47scatterRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter::updates"], [0, 1, 1, "_CPPv47scatterRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter::updates"], [0, 0, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add"], [0, 0, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add::a"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add::a"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add::axes"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add::axis"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add::indices"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add::indices"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add::s"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add::s"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_add::updates"], [0, 1, 1, "_CPPv411scatter_addRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_add::updates"], [0, 0, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max"], [0, 0, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max::a"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max::a"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max::axes"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max::axis"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max::indices"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max::indices"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max::s"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max::s"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_max::updates"], [0, 1, 1, "_CPPv411scatter_maxRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_max::updates"], [0, 0, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min"], [0, 0, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min::a"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min::a"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min::axes"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min::axis"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min::indices"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min::indices"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min::s"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min::s"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_min::updates"], [0, 1, 1, "_CPPv411scatter_minRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_min::updates"], [0, 0, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod"], [0, 0, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod::a"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod::a"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod::axes"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod::axis"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod::indices"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod::indices"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod::s"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod::s"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRK5arrayRK5arrayi14StreamOrDevice", "scatter_prod::updates"], [0, 1, 1, "_CPPv412scatter_prodRK5arrayRKNSt6vectorI5arrayEERK5arrayRKNSt6vectorIiEE14StreamOrDevice", "scatter_prod::updates"], [0, 0, 1, "_CPPv47sigmoidRK5array14StreamOrDevice", "sigmoid"], [0, 1, 1, "_CPPv47sigmoidRK5array14StreamOrDevice", "sigmoid::a"], [0, 1, 1, "_CPPv47sigmoidRK5array14StreamOrDevice", "sigmoid::s"], [0, 0, 1, "_CPPv44signRK5array14StreamOrDevice", "sign"], [0, 1, 1, "_CPPv44signRK5array14StreamOrDevice", "sign::a"], [0, 1, 1, "_CPPv44signRK5array14StreamOrDevice", "sign::s"], [0, 0, 1, "_CPPv43sinRK5array14StreamOrDevice", "sin"], [0, 1, 1, "_CPPv43sinRK5array14StreamOrDevice", "sin::a"], [0, 1, 1, "_CPPv43sinRK5array14StreamOrDevice", "sin::s"], [0, 0, 1, "_CPPv44sinhRK5array14StreamOrDevice", "sinh"], [0, 1, 1, "_CPPv44sinhRK5array14StreamOrDevice", "sinh::a"], [0, 1, 1, "_CPPv44sinhRK5array14StreamOrDevice", "sinh::s"], [0, 0, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice"], [0, 0, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::a"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::a"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::s"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::s"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::start"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::start"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::stop"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::stop"], [0, 1, 1, "_CPPv45sliceRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice::strides"], [0, 0, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update"], [0, 0, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::s"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::s"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::src"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::src"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::start"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::start"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::stop"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::stop"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::strides"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::update"], [0, 1, 1, "_CPPv412slice_updateRK5arrayRK5arrayNSt6vectorIiEENSt6vectorIiEENSt6vectorIiEE14StreamOrDevice", "slice_update::update"], [0, 0, 1, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "softmax"], [0, 0, 1, "_CPPv47softmaxRK5arrayb14StreamOrDevice", "softmax"], [0, 0, 1, "_CPPv47softmaxRK5arrayib14StreamOrDevice", "softmax"], [0, 1, 1, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "softmax::a"], [0, 1, 1, "_CPPv47softmaxRK5arrayb14StreamOrDevice", "softmax::a"], [0, 1, 1, "_CPPv47softmaxRK5arrayib14StreamOrDevice", "softmax::a"], [0, 1, 1, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "softmax::axes"], [0, 1, 1, "_CPPv47softmaxRK5arrayib14StreamOrDevice", "softmax::axis"], [0, 1, 1, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "softmax::precise"], [0, 1, 1, "_CPPv47softmaxRK5arrayb14StreamOrDevice", "softmax::precise"], [0, 1, 1, "_CPPv47softmaxRK5arrayib14StreamOrDevice", "softmax::precise"], [0, 1, 1, "_CPPv47softmaxRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "softmax::s"], [0, 1, 1, "_CPPv47softmaxRK5arrayb14StreamOrDevice", "softmax::s"], [0, 1, 1, "_CPPv47softmaxRK5arrayib14StreamOrDevice", "softmax::s"], [0, 0, 1, "_CPPv44sortRK5array14StreamOrDevice", "sort"], [0, 0, 1, "_CPPv44sortRK5arrayi14StreamOrDevice", "sort"], [0, 1, 1, "_CPPv44sortRK5array14StreamOrDevice", "sort::a"], [0, 1, 1, "_CPPv44sortRK5arrayi14StreamOrDevice", "sort::a"], [0, 1, 1, "_CPPv44sortRK5arrayi14StreamOrDevice", "sort::axis"], [0, 1, 1, "_CPPv44sortRK5array14StreamOrDevice", "sort::s"], [0, 1, 1, "_CPPv44sortRK5arrayi14StreamOrDevice", "sort::s"], [0, 0, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "split"], [0, 0, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "split"], [0, 0, 1, "_CPPv45splitRK5arrayi14StreamOrDevice", "split"], [0, 0, 1, "_CPPv45splitRK5arrayii14StreamOrDevice", "split"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "split::a"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "split::a"], [0, 1, 1, "_CPPv45splitRK5arrayi14StreamOrDevice", "split::a"], [0, 1, 1, "_CPPv45splitRK5arrayii14StreamOrDevice", "split::a"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "split::axis"], [0, 1, 1, "_CPPv45splitRK5arrayii14StreamOrDevice", "split::axis"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "split::indices"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "split::indices"], [0, 1, 1, "_CPPv45splitRK5arrayi14StreamOrDevice", "split::num_splits"], [0, 1, 1, "_CPPv45splitRK5arrayii14StreamOrDevice", "split::num_splits"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "split::s"], [0, 1, 1, "_CPPv45splitRK5arrayRKNSt6vectorIiEEi14StreamOrDevice", "split::s"], [0, 1, 1, "_CPPv45splitRK5arrayi14StreamOrDevice", "split::s"], [0, 1, 1, "_CPPv45splitRK5arrayii14StreamOrDevice", "split::s"], [0, 0, 1, "_CPPv44sqrtRK5array14StreamOrDevice", "sqrt"], [0, 1, 1, "_CPPv44sqrtRK5array14StreamOrDevice", "sqrt::a"], [0, 1, 1, "_CPPv44sqrtRK5array14StreamOrDevice", "sqrt::s"], [0, 0, 1, "_CPPv46squareRK5array14StreamOrDevice", "square"], [0, 1, 1, "_CPPv46squareRK5array14StreamOrDevice", "square::a"], [0, 1, 1, "_CPPv46squareRK5array14StreamOrDevice", "square::s"], [0, 0, 1, "_CPPv47squeezeRK5array14StreamOrDevice", "squeeze"], [0, 0, 1, "_CPPv47squeezeRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "squeeze"], [0, 0, 1, "_CPPv47squeezeRK5arrayi14StreamOrDevice", "squeeze"], [0, 1, 1, "_CPPv47squeezeRK5array14StreamOrDevice", "squeeze::a"], [0, 1, 1, "_CPPv47squeezeRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "squeeze::a"], [0, 1, 1, "_CPPv47squeezeRK5arrayi14StreamOrDevice", "squeeze::a"], [0, 1, 1, "_CPPv47squeezeRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "squeeze::axes"], [0, 1, 1, "_CPPv47squeezeRK5arrayi14StreamOrDevice", "squeeze::axis"], [0, 1, 1, "_CPPv47squeezeRK5array14StreamOrDevice", "squeeze::s"], [0, 1, 1, "_CPPv47squeezeRK5arrayRKNSt6vectorIiEE14StreamOrDevice", "squeeze::s"], [0, 1, 1, "_CPPv47squeezeRK5arrayi14StreamOrDevice", "squeeze::s"], [0, 0, 1, "_CPPv45stackRKNSt6vectorI5arrayEE14StreamOrDevice", "stack"], [0, 0, 1, "_CPPv45stackRKNSt6vectorI5arrayEEi14StreamOrDevice", "stack"], [0, 1, 1, "_CPPv45stackRKNSt6vectorI5arrayEE14StreamOrDevice", "stack::arrays"], [0, 1, 1, "_CPPv45stackRKNSt6vectorI5arrayEEi14StreamOrDevice", "stack::arrays"], [0, 1, 1, "_CPPv45stackRKNSt6vectorI5arrayEEi14StreamOrDevice", "stack::axis"], [0, 1, 1, "_CPPv45stackRKNSt6vectorI5arrayEE14StreamOrDevice", "stack::s"], [0, 1, 1, "_CPPv45stackRKNSt6vectorI5arrayEEi14StreamOrDevice", "stack::s"], [0, 0, 1, "_CPPv4StRK5array14StreamOrDevice", "std"], [0, 0, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std"], [0, 0, 1, "_CPPv4StRK5arraybi14StreamOrDevice", "std"], [0, 0, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std"], [0, 1, 1, "_CPPv4StRK5array14StreamOrDevice", "std::a"], [0, 1, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std::a"], [0, 1, 1, "_CPPv4StRK5arraybi14StreamOrDevice", "std::a"], [0, 1, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std::a"], [0, 1, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std::axes"], [0, 1, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std::axis"], [0, 1, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std::ddof"], [0, 1, 1, "_CPPv4StRK5arraybi14StreamOrDevice", "std::ddof"], [0, 1, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std::ddof"], [0, 1, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std::keepdims"], [0, 1, 1, "_CPPv4StRK5arraybi14StreamOrDevice", "std::keepdims"], [0, 1, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std::keepdims"], [0, 1, 1, "_CPPv4StRK5array14StreamOrDevice", "std::s"], [0, 1, 1, "_CPPv4StRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "std::s"], [0, 1, 1, "_CPPv4StRK5arraybi14StreamOrDevice", "std::s"], [0, 1, 1, "_CPPv4StRK5arrayibi14StreamOrDevice", "std::s"], [0, 0, 1, "_CPPv413stop_gradientRK5array14StreamOrDevice", "stop_gradient"], [0, 1, 1, "_CPPv413stop_gradientRK5array14StreamOrDevice", "stop_gradient::a"], [0, 1, 1, "_CPPv413stop_gradientRK5array14StreamOrDevice", "stop_gradient::s"], [0, 0, 1, "_CPPv48subtractRK5arrayRK5array14StreamOrDevice", "subtract"], [0, 1, 1, "_CPPv48subtractRK5arrayRK5array14StreamOrDevice", "subtract::a"], [0, 1, 1, "_CPPv48subtractRK5arrayRK5array14StreamOrDevice", "subtract::b"], [0, 1, 1, "_CPPv48subtractRK5arrayRK5array14StreamOrDevice", "subtract::s"], [0, 0, 1, "_CPPv43sumRK5array14StreamOrDevice", "sum"], [0, 0, 1, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "sum"], [0, 0, 1, "_CPPv43sumRK5arrayb14StreamOrDevice", "sum"], [0, 0, 1, "_CPPv43sumRK5arrayib14StreamOrDevice", "sum"], [0, 1, 1, "_CPPv43sumRK5array14StreamOrDevice", "sum::a"], [0, 1, 1, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "sum::a"], [0, 1, 1, "_CPPv43sumRK5arrayb14StreamOrDevice", "sum::a"], [0, 1, 1, "_CPPv43sumRK5arrayib14StreamOrDevice", "sum::a"], [0, 1, 1, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "sum::axes"], [0, 1, 1, "_CPPv43sumRK5arrayib14StreamOrDevice", "sum::axis"], [0, 1, 1, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "sum::keepdims"], [0, 1, 1, "_CPPv43sumRK5arrayb14StreamOrDevice", "sum::keepdims"], [0, 1, 1, "_CPPv43sumRK5arrayib14StreamOrDevice", "sum::keepdims"], [0, 1, 1, "_CPPv43sumRK5array14StreamOrDevice", "sum::s"], [0, 1, 1, "_CPPv43sumRK5arrayRKNSt6vectorIiEEb14StreamOrDevice", "sum::s"], [0, 1, 1, "_CPPv43sumRK5arrayb14StreamOrDevice", "sum::s"], [0, 1, 1, "_CPPv43sumRK5arrayib14StreamOrDevice", "sum::s"], [0, 0, 1, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", "swapaxes"], [0, 1, 1, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", "swapaxes::a"], [0, 1, 1, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", "swapaxes::axis1"], [0, 1, 1, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", "swapaxes::axis2"], [0, 1, 1, "_CPPv48swapaxesRK5arrayii14StreamOrDevice", "swapaxes::s"], [0, 0, 1, "_CPPv44takeRK5arrayRK5array14StreamOrDevice", "take"], [0, 0, 1, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", "take"], [0, 0, 1, "_CPPv44takeRK5arrayi14StreamOrDevice", "take"], [0, 0, 1, "_CPPv44takeRK5arrayii14StreamOrDevice", "take"], [0, 1, 1, "_CPPv44takeRK5arrayRK5array14StreamOrDevice", "take::a"], [0, 1, 1, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", "take::a"], [0, 1, 1, "_CPPv44takeRK5arrayi14StreamOrDevice", "take::a"], [0, 1, 1, "_CPPv44takeRK5arrayii14StreamOrDevice", "take::a"], [0, 1, 1, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", "take::axis"], [0, 1, 1, "_CPPv44takeRK5arrayii14StreamOrDevice", "take::axis"], [0, 1, 1, "_CPPv44takeRK5arrayi14StreamOrDevice", "take::index"], [0, 1, 1, "_CPPv44takeRK5arrayii14StreamOrDevice", "take::index"], [0, 1, 1, "_CPPv44takeRK5arrayRK5array14StreamOrDevice", "take::indices"], [0, 1, 1, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", "take::indices"], [0, 1, 1, "_CPPv44takeRK5arrayRK5array14StreamOrDevice", "take::s"], [0, 1, 1, "_CPPv44takeRK5arrayRK5arrayi14StreamOrDevice", "take::s"], [0, 1, 1, "_CPPv44takeRK5arrayi14StreamOrDevice", "take::s"], [0, 1, 1, "_CPPv44takeRK5arrayii14StreamOrDevice", "take::s"], [0, 0, 1, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", "take_along_axis"], [0, 1, 1, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", "take_along_axis::a"], [0, 1, 1, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", "take_along_axis::axis"], [0, 1, 1, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", "take_along_axis::indices"], [0, 1, 1, "_CPPv415take_along_axisRK5arrayRK5arrayi14StreamOrDevice", "take_along_axis::s"], [0, 0, 1, "_CPPv43tanRK5array14StreamOrDevice", "tan"], [0, 1, 1, "_CPPv43tanRK5array14StreamOrDevice", "tan::a"], [0, 1, 1, "_CPPv43tanRK5array14StreamOrDevice", "tan::s"], [0, 0, 1, "_CPPv44tanhRK5array14StreamOrDevice", "tanh"], [0, 1, 1, "_CPPv44tanhRK5array14StreamOrDevice", "tanh::a"], [0, 1, 1, "_CPPv44tanhRK5array14StreamOrDevice", "tanh::s"], [0, 0, 1, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", "tensordot"], [0, 0, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", "tensordot::a"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot::a"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot::axes_a"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot::axes_b"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", "tensordot::axis"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", "tensordot::b"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot::b"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayKi14StreamOrDevice", "tensordot::s"], [0, 1, 1, "_CPPv49tensordotRK5arrayRK5arrayRKNSt6vectorIiEERKNSt6vectorIiEE14StreamOrDevice", "tensordot::s"], [0, 0, 1, "_CPPv44tileRK5arrayNSt6vectorIiEE14StreamOrDevice", "tile"], [0, 1, 1, "_CPPv44tileRK5arrayNSt6vectorIiEE14StreamOrDevice", "tile::arr"], [0, 1, 1, "_CPPv44tileRK5arrayNSt6vectorIiEE14StreamOrDevice", "tile::reps"], [0, 1, 1, "_CPPv44tileRK5arrayNSt6vectorIiEE14StreamOrDevice", "tile::s"], [0, 0, 1, "_CPPv44topkRK5arrayi14StreamOrDevice", "topk"], [0, 0, 1, "_CPPv44topkRK5arrayii14StreamOrDevice", "topk"], [0, 1, 1, "_CPPv44topkRK5arrayi14StreamOrDevice", "topk::a"], [0, 1, 1, "_CPPv44topkRK5arrayii14StreamOrDevice", "topk::a"], [0, 1, 1, "_CPPv44topkRK5arrayii14StreamOrDevice", "topk::axis"], [0, 1, 1, "_CPPv44topkRK5arrayi14StreamOrDevice", "topk::k"], [0, 1, 1, "_CPPv44topkRK5arrayii14StreamOrDevice", "topk::k"], [0, 1, 1, "_CPPv44topkRK5arrayi14StreamOrDevice", "topk::s"], [0, 1, 1, "_CPPv44topkRK5arrayii14StreamOrDevice", "topk::s"], [0, 0, 1, "_CPPv45traceRK5array14StreamOrDevice", "trace"], [0, 0, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace"], [0, 0, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace"], [0, 1, 1, "_CPPv45traceRK5array14StreamOrDevice", "trace::a"], [0, 1, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace::a"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::a"], [0, 1, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace::axis1"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::axis1"], [0, 1, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace::axis2"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::axis2"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::dtype"], [0, 1, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace::offset"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::offset"], [0, 1, 1, "_CPPv45traceRK5array14StreamOrDevice", "trace::s"], [0, 1, 1, "_CPPv45traceRK5arrayiii14StreamOrDevice", "trace::s"], [0, 1, 1, "_CPPv45traceRK5arrayiii5Dtype14StreamOrDevice", "trace::s"], [0, 0, 1, "_CPPv49transposeRK5array14StreamOrDevice", "transpose"], [0, 0, 1, "_CPPv49transposeRK5arrayNSt16initializer_listIiEE14StreamOrDevice", "transpose"], [0, 0, 1, "_CPPv49transposeRK5arrayNSt6vectorIiEE14StreamOrDevice", "transpose"], [0, 1, 1, "_CPPv49transposeRK5array14StreamOrDevice", "transpose::a"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt16initializer_listIiEE14StreamOrDevice", "transpose::a"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt6vectorIiEE14StreamOrDevice", "transpose::a"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt16initializer_listIiEE14StreamOrDevice", "transpose::axes"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt6vectorIiEE14StreamOrDevice", "transpose::axes"], [0, 1, 1, "_CPPv49transposeRK5array14StreamOrDevice", "transpose::s"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt16initializer_listIiEE14StreamOrDevice", "transpose::s"], [0, 1, 1, "_CPPv49transposeRK5arrayNSt6vectorIiEE14StreamOrDevice", "transpose::s"], [0, 0, 1, "_CPPv43trii5Dtype14StreamOrDevice", "tri"], [0, 0, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri"], [0, 1, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri::k"], [0, 1, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri::m"], [0, 1, 1, "_CPPv43trii5Dtype14StreamOrDevice", "tri::n"], [0, 1, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri::n"], [0, 1, 1, "_CPPv43trii5Dtype14StreamOrDevice", "tri::s"], [0, 1, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri::s"], [0, 1, 1, "_CPPv43trii5Dtype14StreamOrDevice", "tri::type"], [0, 1, 1, "_CPPv43triiii5Dtype14StreamOrDevice", "tri::type"], [0, 0, 1, "_CPPv44tril5arrayi14StreamOrDevice", "tril"], [0, 1, 1, "_CPPv44tril5arrayi14StreamOrDevice", "tril::k"], [0, 1, 1, "_CPPv44tril5arrayi14StreamOrDevice", "tril::s"], [0, 1, 1, "_CPPv44tril5arrayi14StreamOrDevice", "tril::x"], [0, 0, 1, "_CPPv44triu5arrayi14StreamOrDevice", "triu"], [0, 1, 1, "_CPPv44triu5arrayi14StreamOrDevice", "triu::k"], [0, 1, 1, "_CPPv44triu5arrayi14StreamOrDevice", "triu::s"], [0, 1, 1, "_CPPv44triu5arrayi14StreamOrDevice", "triu::x"], [0, 0, 1, "_CPPv43varRK5array14StreamOrDevice", "var"], [0, 0, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var"], [0, 0, 1, "_CPPv43varRK5arraybi14StreamOrDevice", "var"], [0, 0, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var"], [0, 1, 1, "_CPPv43varRK5array14StreamOrDevice", "var::a"], [0, 1, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var::a"], [0, 1, 1, "_CPPv43varRK5arraybi14StreamOrDevice", "var::a"], [0, 1, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var::a"], [0, 1, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var::axes"], [0, 1, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var::axis"], [0, 1, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var::ddof"], [0, 1, 1, "_CPPv43varRK5arraybi14StreamOrDevice", "var::ddof"], [0, 1, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var::ddof"], [0, 1, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var::keepdims"], [0, 1, 1, "_CPPv43varRK5arraybi14StreamOrDevice", "var::keepdims"], [0, 1, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var::keepdims"], [0, 1, 1, "_CPPv43varRK5array14StreamOrDevice", "var::s"], [0, 1, 1, "_CPPv43varRK5arrayRKNSt6vectorIiEEbi14StreamOrDevice", "var::s"], [0, 1, 1, "_CPPv43varRK5arraybi14StreamOrDevice", "var::s"], [0, 1, 1, "_CPPv43varRK5arrayibi14StreamOrDevice", "var::s"], [0, 0, 1, "_CPPv44viewRK5arrayRK5Dtype14StreamOrDevice", "view"], [0, 1, 1, "_CPPv44viewRK5arrayRK5Dtype14StreamOrDevice", "view::a"], [0, 1, 1, "_CPPv44viewRK5arrayRK5Dtype14StreamOrDevice", "view::dtype"], [0, 1, 1, "_CPPv44viewRK5arrayRK5Dtype14StreamOrDevice", "view::s"], [0, 0, 1, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", "where"], [0, 1, 1, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", "where::condition"], [0, 1, 1, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", "where::s"], [0, 1, 1, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", "where::x"], [0, 1, 1, "_CPPv45whereRK5arrayRK5arrayRK5array14StreamOrDevice", "where::y"], [0, 0, 1, "_CPPv45zerosRKNSt6vectorIiEE14StreamOrDevice", "zeros"], [0, 0, 1, "_CPPv45zerosRKNSt6vectorIiEE5Dtype14StreamOrDevice", "zeros"], [0, 1, 1, "_CPPv45zerosRKNSt6vectorIiEE5Dtype14StreamOrDevice", "zeros::dtype"], [0, 1, 1, "_CPPv45zerosRKNSt6vectorIiEE14StreamOrDevice", "zeros::s"], [0, 1, 1, "_CPPv45zerosRKNSt6vectorIiEE5Dtype14StreamOrDevice", "zeros::s"], [0, 1, 1, "_CPPv45zerosRKNSt6vectorIiEE14StreamOrDevice", "zeros::shape"], [0, 1, 1, "_CPPv45zerosRKNSt6vectorIiEE5Dtype14StreamOrDevice", "zeros::shape"], [0, 0, 1, "_CPPv410zeros_likeRK5array14StreamOrDevice", "zeros_like"], [0, 1, 1, "_CPPv410zeros_likeRK5array14StreamOrDevice", "zeros_like::a"], [0, 1, 1, "_CPPv410zeros_likeRK5array14StreamOrDevice", "zeros_like::s"]], "mlx.core": [[9, 3, 1, "", "Device"], [10, 3, 1, "", "Dtype"], [11, 3, 1, "", "DtypeCategory"], [315, 3, 1, "", "Stream"], [12, 5, 1, "", "abs"], [13, 5, 1, "", "add"], [14, 5, 1, "", "addmm"], [15, 5, 1, "", "all"], [16, 5, 1, "", "allclose"], [17, 5, 1, "", "any"], [18, 5, 1, "", "arange"], [19, 5, 1, "", "arccos"], [20, 5, 1, "", "arccosh"], [21, 5, 1, "", "arcsin"], [22, 5, 1, "", "arcsinh"], [23, 5, 1, "", "arctan"], [24, 5, 1, "", "arctan2"], [25, 5, 1, "", "arctanh"], [26, 5, 1, "", "argmax"], [27, 5, 1, "", "argmin"], [28, 5, 1, "", "argpartition"], [29, 5, 1, "", "argsort"], [30, 3, 1, "", "array"], [82, 5, 1, "", "array_equal"], [83, 5, 1, "", "as_strided"], [84, 5, 1, "", "atleast_1d"], [85, 5, 1, "", "atleast_2d"], [86, 5, 1, "", "atleast_3d"], [87, 5, 1, "", "bitwise_and"], [88, 5, 1, "", "bitwise_or"], [89, 5, 1, "", "bitwise_xor"], [90, 5, 1, "", "block_masked_mm"], [91, 5, 1, "", "broadcast_to"], [92, 5, 1, "", "ceil"], [93, 5, 1, "", "clip"], [94, 5, 1, "", "compile"], [95, 5, 1, "", "concatenate"], [96, 5, 1, "", "conj"], [97, 5, 1, "", "conjugate"], [98, 5, 1, "", "conv1d"], [99, 5, 1, "", "conv2d"], [100, 5, 1, "", "conv3d"], [101, 5, 1, "", "conv_general"], [102, 5, 1, "", "conv_transpose1d"], [103, 5, 1, "", "conv_transpose2d"], [104, 5, 1, "", "conv_transpose3d"], [105, 5, 1, "", "convolve"], [106, 5, 1, "", "cos"], [107, 5, 1, "", "cosh"], [108, 5, 1, "", "cummax"], [109, 5, 1, "", "cummin"], [110, 5, 1, "", "cumprod"], [111, 5, 1, "", "cumsum"], [112, 3, 1, "", "custom_function"], [113, 5, 1, "", "default_device"], [114, 5, 1, "", "default_stream"], [115, 5, 1, "", "degrees"], [116, 5, 1, "", "dequantize"], [117, 5, 1, "", "diag"], [118, 5, 1, "", "diagonal"], [119, 5, 1, "", "disable_compile"], [128, 5, 1, "", "divide"], [129, 5, 1, "", "divmod"], [130, 5, 1, "", "einsum"], [131, 5, 1, "", "einsum_path"], [132, 5, 1, "", "enable_compile"], [133, 5, 1, "", "equal"], [134, 5, 1, "", "erf"], [135, 5, 1, "", "erfinv"], [136, 5, 1, "", "eval"], [137, 5, 1, "", "exp"], [138, 5, 1, "", "expand_dims"], [139, 5, 1, "", "expm1"], [140, 5, 1, "", "eye"], [159, 5, 1, "", "flatten"], [160, 5, 1, "", "floor"], [161, 5, 1, "", "floor_divide"], [162, 5, 1, "", "full"], [163, 5, 1, "", "gather_mm"], [164, 5, 1, "", "gather_qmm"], [165, 5, 1, "", "grad"], [166, 5, 1, "", "greater"], [167, 5, 1, "", "greater_equal"], [168, 5, 1, "", "hadamard_transform"], [169, 5, 1, "", "identity"], [170, 5, 1, "", "imag"], [171, 5, 1, "", "inner"], [172, 5, 1, "", "isclose"], [173, 5, 1, "", "isfinite"], [174, 5, 1, "", "isinf"], [175, 5, 1, "", "isnan"], [176, 5, 1, "", "isneginf"], [177, 5, 1, "", "isposinf"], [178, 5, 1, "", "issubdtype"], [179, 5, 1, "", "jvp"], [180, 5, 1, "", "left_shift"], [181, 5, 1, "", "less"], [182, 5, 1, "", "less_equal"], [193, 5, 1, "", "linspace"], [194, 5, 1, "", "load"], [195, 5, 1, "", "log"], [196, 5, 1, "", "log10"], [197, 5, 1, "", "log1p"], [198, 5, 1, "", "log2"], [199, 5, 1, "", "logaddexp"], [200, 5, 1, "", "logical_and"], [201, 5, 1, "", "logical_not"], [202, 5, 1, "", "logical_or"], [203, 5, 1, "", "logsumexp"], [204, 5, 1, "", "matmul"], [205, 5, 1, "", "max"], [206, 5, 1, "", "maximum"], [207, 5, 1, "", "mean"], [208, 5, 1, "", "meshgrid"], [221, 5, 1, "", "min"], [222, 5, 1, "", "minimum"], [223, 5, 1, "", "moveaxis"], [224, 5, 1, "", "multiply"], [225, 5, 1, "", "nan_to_num"], [226, 5, 1, "", "negative"], [227, 5, 1, "", "new_stream"], [228, 5, 1, "", "not_equal"], [229, 5, 1, "", "ones"], [230, 5, 1, "", "ones_like"], [231, 5, 1, "", "outer"], [232, 5, 1, "", "pad"], [233, 5, 1, "", "partition"], [234, 5, 1, "", "power"], [235, 5, 1, "", "prod"], [236, 5, 1, "", "put_along_axis"], [237, 5, 1, "", "quantize"], [238, 5, 1, "", "quantized_matmul"], [239, 5, 1, "", "radians"], [253, 5, 1, "", "real"], [254, 5, 1, "", "reciprocal"], [255, 5, 1, "", "remainder"], [256, 5, 1, "", "repeat"], [257, 5, 1, "", "reshape"], [258, 5, 1, "", "right_shift"], [259, 5, 1, "", "roll"], [260, 5, 1, "", "round"], [261, 5, 1, "", "rsqrt"], [262, 5, 1, "", "save"], [263, 5, 1, "", "save_gguf"], [264, 5, 1, "", "save_safetensors"], [265, 5, 1, "", "savez"], [266, 5, 1, "", "savez_compressed"], [267, 5, 1, "", "set_default_device"], [268, 5, 1, "", "set_default_stream"], [269, 5, 1, "", "sigmoid"], [270, 5, 1, "", "sign"], [271, 5, 1, "", "sin"], [272, 5, 1, "", "sinh"], [273, 5, 1, "", "softmax"], [274, 5, 1, "", "sort"], [275, 5, 1, "", "split"], [276, 5, 1, "", "sqrt"], [277, 5, 1, "", "square"], [278, 5, 1, "", "squeeze"], [279, 5, 1, "", "stack"], [280, 5, 1, "", "std"], [281, 5, 1, "", "stop_gradient"], [282, 5, 1, "", "stream"], [283, 5, 1, "", "subtract"], [284, 5, 1, "", "sum"], [285, 5, 1, "", "swapaxes"], [286, 5, 1, "", "synchronize"], [287, 5, 1, "", "take"], [288, 5, 1, "", "take_along_axis"], [289, 5, 1, "", "tan"], [290, 5, 1, "", "tanh"], [291, 5, 1, "", "tensordot"], [292, 5, 1, "", "tile"], [293, 5, 1, "", "topk"], [294, 5, 1, "", "trace"], [295, 5, 1, "", "transpose"], [296, 5, 1, "", "tri"], [297, 5, 1, "", "tril"], [298, 5, 1, "", "triu"], [299, 5, 1, "", "value_and_grad"], [300, 5, 1, "", "var"], [301, 5, 1, "", "view"], [302, 5, 1, "", "vjp"], [303, 5, 1, "", "vmap"], [304, 5, 1, "", "where"], [305, 5, 1, "", "zeros"], [306, 5, 1, "", "zeros_like"]], "mlx.core.Device": [[9, 4, 1, "", "__init__"]], "mlx.core.Dtype": [[10, 4, 1, "", "__init__"]], "mlx.core.DtypeCategory": [[11, 4, 1, "", "__init__"]], "mlx.core.Stream": [[315, 4, 1, "", "__init__"]], "mlx.core.array": [[31, 6, 1, "", "T"], [30, 4, 1, "", "__init__"], [32, 4, 1, "", "abs"], [33, 4, 1, "", "all"], [34, 4, 1, "", "any"], [35, 4, 1, "", "argmax"], [36, 4, 1, "", "argmin"], [37, 4, 1, "", "astype"], [38, 6, 1, "", "at"], [39, 4, 1, "", "conj"], [40, 4, 1, "", "cos"], [41, 4, 1, "", "cummax"], [42, 4, 1, "", "cummin"], [43, 4, 1, "", "cumprod"], [44, 4, 1, "", "cumsum"], [45, 4, 1, "", "diag"], [46, 4, 1, "", "diagonal"], [47, 6, 1, "", "dtype"], [48, 4, 1, "", "exp"], [49, 4, 1, "", "flatten"], [50, 4, 1, "", "item"], [51, 6, 1, "", "itemsize"], [52, 4, 1, "", "log"], [53, 4, 1, "", "log10"], [54, 4, 1, "", "log1p"], [55, 4, 1, "", "log2"], [56, 4, 1, "", "logsumexp"], [57, 4, 1, "", "max"], [58, 4, 1, "", "mean"], [59, 4, 1, "", "min"], [60, 4, 1, "", "moveaxis"], [61, 6, 1, "", "nbytes"], [62, 6, 1, "", "ndim"], [63, 4, 1, "", "prod"], [64, 4, 1, "", "reciprocal"], [65, 4, 1, "", "reshape"], [66, 4, 1, "", "round"], [67, 4, 1, "", "rsqrt"], [68, 6, 1, "", "shape"], [69, 4, 1, "", "sin"], [70, 6, 1, "", "size"], [71, 4, 1, "", "split"], [72, 4, 1, "", "sqrt"], [73, 4, 1, "", "square"], [74, 4, 1, "", "squeeze"], [75, 4, 1, "", "std"], [76, 4, 1, "", "sum"], [77, 4, 1, "", "swapaxes"], [78, 4, 1, "", "tolist"], [79, 4, 1, "", "transpose"], [80, 4, 1, "", "var"], [81, 4, 1, "", "view"]], "mlx.core.custom_function": [[112, 4, 1, "", "__init__"]], "mlx.core.distributed": [[120, 3, 1, "", "Group"], [121, 5, 1, "", "all_gather"], [122, 5, 1, "", "all_sum"], [123, 5, 1, "", "init"], [124, 5, 1, "", "is_available"], [125, 5, 1, "", "recv"], [126, 5, 1, "", "recv_like"], [127, 5, 1, "", "send"]], "mlx.core.distributed.Group": [[120, 4, 1, "", "__init__"]], "mlx.core.fast": [[141, 5, 1, "", "affine_quantize"], [142, 5, 1, "", "layer_norm"], [143, 5, 1, "", "metal_kernel"], [144, 5, 1, "", "rms_norm"], [145, 5, 1, "", "rope"], [146, 5, 1, "", "scaled_dot_product_attention"]], "mlx.core.fft": [[147, 5, 1, "", "fft"], [148, 5, 1, "", "fft2"], [149, 5, 1, "", "fftn"], [150, 5, 1, "", "ifft"], [151, 5, 1, "", "ifft2"], [152, 5, 1, "", "ifftn"], [153, 5, 1, "", "irfft"], [154, 5, 1, "", "irfft2"], [155, 5, 1, "", "irfftn"], [156, 5, 1, "", "rfft"], [157, 5, 1, "", "rfft2"], [158, 5, 1, "", "rfftn"]], "mlx.core.linalg": [[183, 5, 1, "", "cholesky"], [184, 5, 1, "", "cholesky_inv"], [185, 5, 1, "", "cross"], [186, 5, 1, "", "eigh"], [187, 5, 1, "", "eigvalsh"], [188, 5, 1, "", "inv"], [189, 5, 1, "", "norm"], [190, 5, 1, "", "qr"], [191, 5, 1, "", "svd"], [192, 5, 1, "", "tri_inv"]], "mlx.core.metal": [[209, 5, 1, "", "clear_cache"], [210, 5, 1, "", "device_info"], [211, 5, 1, "", "get_active_memory"], [212, 5, 1, "", "get_cache_memory"], [213, 5, 1, "", "get_peak_memory"], [214, 5, 1, "", "is_available"], [215, 5, 1, "", "reset_peak_memory"], [216, 5, 1, "", "set_cache_limit"], [217, 5, 1, "", "set_memory_limit"], [218, 5, 1, "", "set_wired_limit"], [219, 5, 1, "", "start_capture"], [220, 5, 1, "", "stop_capture"]], "mlx.core.random": [[240, 5, 1, "", "bernoulli"], [241, 5, 1, "", "categorical"], [242, 5, 1, "", "gumbel"], [243, 5, 1, "", "key"], [244, 5, 1, "", "laplace"], [245, 5, 1, "", "multivariate_normal"], [246, 5, 1, "", "normal"], [247, 5, 1, "", "permutation"], [248, 5, 1, "", "randint"], [249, 5, 1, "", "seed"], [250, 5, 1, "", "split"], [251, 5, 1, "", "truncated_normal"], [252, 5, 1, "", "uniform"]], "mlx.nn": [[325, 3, 1, "", "ALiBi"], [326, 3, 1, "", "AvgPool1d"], [327, 3, 1, "", "AvgPool2d"], [328, 3, 1, "", "BatchNorm"], [329, 3, 1, "", "CELU"], [330, 3, 1, "", "Conv1d"], [331, 3, 1, "", "Conv2d"], [332, 3, 1, "", "Conv3d"], [333, 3, 1, "", "ConvTranspose1d"], [334, 3, 1, "", "ConvTranspose2d"], [335, 3, 1, "", "ConvTranspose3d"], [336, 3, 1, "", "Dropout"], [337, 3, 1, "", "Dropout2d"], [338, 3, 1, "", "Dropout3d"], [339, 3, 1, "", "ELU"], [340, 3, 1, "", "Embedding"], [341, 3, 1, "", "GELU"], [342, 3, 1, "", "GLU"], [343, 3, 1, "", "GRU"], [344, 3, 1, "", "GroupNorm"], [345, 3, 1, "", "HardShrink"], [346, 3, 1, "", "HardTanh"], [347, 3, 1, "", "Hardswish"], [348, 3, 1, "", "InstanceNorm"], [349, 3, 1, "", "LSTM"], [350, 3, 1, "", "LayerNorm"], [351, 3, 1, "", "LeakyReLU"], [352, 3, 1, "", "Linear"], [353, 3, 1, "", "LogSigmoid"], [354, 3, 1, "", "LogSoftmax"], [355, 3, 1, "", "MaxPool1d"], [356, 3, 1, "", "MaxPool2d"], [357, 3, 1, "", "Mish"], [452, 3, 1, "", "Module"], [378, 3, 1, "", "MultiHeadAttention"], [379, 3, 1, "", "PReLU"], [380, 3, 1, "", "QuantizedEmbedding"], [381, 3, 1, "", "QuantizedLinear"], [382, 3, 1, "", "RMSNorm"], [383, 3, 1, "", "RNN"], [384, 3, 1, "", "ReLU"], [385, 3, 1, "", "ReLU6"], [386, 3, 1, "", "RoPE"], [387, 3, 1, "", "SELU"], [388, 3, 1, "", "Sequential"], [389, 3, 1, "", "SiLU"], [390, 3, 1, "", "Sigmoid"], [391, 3, 1, "", "SinusoidalPositionalEncoding"], [392, 3, 1, "", "Softmax"], [393, 3, 1, "", "Softmin"], [394, 3, 1, "", "Softplus"], [395, 3, 1, "", "Softshrink"], [396, 3, 1, "", "Softsign"], [397, 3, 1, "", "Step"], [398, 3, 1, "", "Tanh"], [399, 3, 1, "", "Transformer"], [400, 3, 1, "", "Upsample"], [409, 3, 1, "", "celu"], [410, 3, 1, "", "elu"], [411, 3, 1, "", "gelu"], [412, 3, 1, "", "gelu_approx"], [413, 3, 1, "", "gelu_fast_approx"], [414, 3, 1, "", "glu"], [415, 3, 1, "", "hard_shrink"], [416, 3, 1, "", "hard_tanh"], [417, 3, 1, "", "hardswish"], [418, 3, 1, "", "leaky_relu"], [419, 3, 1, "", "log_sigmoid"], [420, 3, 1, "", "log_softmax"], [435, 3, 1, "", "mish"], [436, 3, 1, "", "prelu"], [307, 5, 1, "", "quantize"], [437, 3, 1, "", "relu"], [438, 3, 1, "", "relu6"], [439, 3, 1, "", "selu"], [440, 3, 1, "", "sigmoid"], [441, 3, 1, "", "silu"], [442, 3, 1, "", "softmax"], [443, 3, 1, "", "softmin"], [444, 3, 1, "", "softplus"], [445, 3, 1, "", "softshrink"], [446, 3, 1, "", "step"], [447, 3, 1, "", "tanh"], [308, 5, 1, "", "value_and_grad"]], "mlx.nn.Module": [[358, 4, 1, "", "apply"], [359, 4, 1, "", "apply_to_modules"], [360, 4, 1, "", "children"], [361, 4, 1, "", "eval"], [362, 4, 1, "", "filter_and_map"], [363, 4, 1, "", "freeze"], [364, 4, 1, "", "leaf_modules"], [365, 4, 1, "", "load_weights"], [366, 4, 1, "", "modules"], [367, 4, 1, "", "named_modules"], [368, 4, 1, "", "parameters"], [369, 4, 1, "", "save_weights"], [370, 4, 1, "", "set_dtype"], [371, 6, 1, "", "state"], [372, 4, 1, "", "train"], [373, 4, 1, "", "trainable_parameters"], [374, 6, 1, "", "training"], [375, 4, 1, "", "unfreeze"], [376, 4, 1, "", "update"], [377, 4, 1, "", "update_modules"]], "mlx.nn.init": [[401, 5, 1, "", "constant"], [402, 5, 1, "", "glorot_normal"], [403, 5, 1, "", "glorot_uniform"], [404, 5, 1, "", "he_normal"], [405, 5, 1, "", "he_uniform"], [406, 5, 1, "", "identity"], [407, 5, 1, "", "normal"], [408, 5, 1, "", "uniform"]], "mlx.nn.losses": [[421, 3, 1, "", "binary_cross_entropy"], [422, 3, 1, "", "cosine_similarity_loss"], [423, 3, 1, "", "cross_entropy"], [424, 3, 1, "", "gaussian_nll_loss"], [425, 3, 1, "", "hinge_loss"], [426, 3, 1, "", "huber_loss"], [427, 3, 1, "", "kl_div_loss"], [428, 3, 1, "", "l1_loss"], [429, 3, 1, "", "log_cosh_loss"], [430, 3, 1, "", "margin_ranking_loss"], [431, 3, 1, "", "mse_loss"], [432, 3, 1, "", "nll_loss"], [433, 3, 1, "", "smooth_l1_loss"], [434, 3, 1, "", "triplet_loss"]], "mlx.optimizers": [[455, 3, 1, "", "AdaDelta"], [456, 3, 1, "", "Adafactor"], [457, 3, 1, "", "Adagrad"], [458, 3, 1, "", "Adam"], [459, 3, 1, "", "AdamW"], [460, 3, 1, "", "Adamax"], [461, 3, 1, "", "Lion"], [474, 3, 1, "", "Optimizer"], [466, 3, 1, "", "RMSprop"], [467, 3, 1, "", "SGD"], [309, 5, 1, "", "clip_grad_norm"], [468, 5, 1, "", "cosine_decay"], [469, 5, 1, "", "exponential_decay"], [470, 5, 1, "", "join_schedules"], [471, 5, 1, "", "linear_schedule"], [472, 5, 1, "", "step_decay"]], "mlx.optimizers.Optimizer": [[462, 4, 1, "", "apply_gradients"], [463, 4, 1, "", "init"], [464, 6, 1, "", "state"], [465, 4, 1, "", "update"]], "mlx.utils": [[310, 5, 1, "", "tree_flatten"], [311, 5, 1, "", "tree_map"], [312, 5, 1, "", "tree_map_with_path"], [313, 5, 1, "", "tree_reduce"], [314, 5, 1, "", "tree_unflatten"]]}, "objnames": {"0": ["cpp", "function", "C++ function"], "1": ["cpp", "functionParam", "C++ function parameter"], "2": ["cpp", "templateParam", "C++ template parameter"], "3": ["py", "class", "Python class"], "4": ["py", "method", "Python method"], "5": ["py", "function", "Python function"], "6": ["py", "property", "Python property"]}, "objtypes": {"0": "cpp:function", "1": "cpp:functionParam", "2": "cpp:templateParam", "3": "py:class", "4": "py:method", "5": "py:function", "6": "py:property"}, "terms": {"": [0, 1, 2, 4, 5, 6, 47, 51, 62, 94, 114, 116, 141, 148, 149, 151, 152, 154, 155, 157, 158, 165, 184, 189, 191, 194, 207, 231, 237, 241, 260, 263, 264, 280, 282, 299, 300, 301, 303, 308, 324, 327, 343, 349, 356, 362, 363, 365, 369, 370, 371, 375, 383, 454, 463, 464, 476, 479, 481, 484, 485, 486, 487], "0": [0, 1, 2, 4, 5, 6, 8, 9, 14, 18, 38, 45, 46, 49, 66, 71, 75, 80, 83, 95, 98, 99, 100, 101, 102, 103, 104, 117, 118, 140, 143, 146, 159, 163, 165, 186, 188, 189, 190, 192, 209, 216, 218, 225, 232, 240, 244, 246, 247, 252, 256, 260, 275, 279, 280, 294, 296, 297, 298, 299, 300, 303, 309, 310, 312, 313, 324, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 344, 345, 348, 350, 351, 355, 356, 379, 384, 386, 391, 395, 397, 399, 401, 402, 403, 404, 405, 406, 407, 408, 409, 410, 412, 413, 415, 416, 417, 418, 421, 423, 425, 426, 430, 433, 434, 436, 437, 438, 439, 445, 446, 449, 452, 455, 456, 458, 459, 460, 461, 463, 466, 467, 468, 469, 470, 471, 472, 476, 479, 480, 481, 482, 483, 484, 485, 486], "00005": 4, "0001": 391, "0005": 412, "001": 456, "00364": 4, "01": [4, 351, 418, 459], "0137595": 404, "015": 413, "0184009": 405, "02264": 403, "025": 481, "02765": 404, "0300242": 405, "044715": [341, 412], "0485873": 423, "05": [16, 172, 328, 344, 348, 350, 382], "0507": 439, "05202": 5, "06": [424, 434, 455], "0638": 430, "06450": 350, "0645099": 407, "06561": 469, "06675": 461, "07467": 382, "08": [16, 172, 422, 457, 458, 459, 460, 466], "08022": 348, "081": 472, "08415": 413, "08494": 344, "08619": 405, "08681": [357, 435], "09864": 5, "0999938": 470, "0999961": 468, "0f": 0, "1": [0, 1, 2, 3, 5, 6, 14, 18, 28, 29, 38, 46, 49, 98, 99, 100, 101, 102, 103, 104, 117, 118, 139, 143, 146, 147, 148, 150, 151, 153, 154, 155, 156, 157, 158, 159, 168, 171, 178, 184, 185, 186, 187, 189, 190, 204, 208, 217, 231, 233, 237, 241, 244, 245, 246, 252, 269, 274, 287, 293, 294, 299, 309, 312, 313, 317, 324, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 342, 343, 344, 348, 349, 350, 352, 355, 356, 379, 382, 383, 386, 390, 391, 397, 400, 402, 403, 404, 405, 406, 407, 408, 409, 410, 412, 413, 414, 416, 419, 420, 421, 422, 423, 424, 425, 426, 427, 429, 430, 432, 433, 434, 439, 440, 442, 443, 444, 446, 449, 452, 454, 455, 456, 457, 458, 459, 460, 461, 463, 466, 467, 468, 469, 470, 471, 472, 479, 480, 481, 482, 484, 485, 486, 487], "10": [0, 3, 5, 6, 196, 260, 265, 311, 324, 365, 449, 470, 472, 479, 480, 482], "100": [2, 4, 5, 421, 471, 479, 481, 483, 487], "1000": [468, 479], "10000": 386, "101": 471, "1024": [1, 5], "105361": 421, "109": 2, "10_000": 4, "10x": 461, "11": 189, "114": 2, "12": [5, 168, 470], "1212": 455, "12451": 403, "128": [265, 324], "13": 8, "14": 8, "15": [1, 8, 189, 218, 313, 479], "150594": 402, "15268": 404, "16": [1, 143, 317, 326, 348, 355, 358, 452], "1606": 413, "1607": [348, 350], "16384": 168, "16506": 405, "17": 8, "177208": 404, "1803": 344, "1908": [357, 435], "1910": 382, "191107": 402, "1985": 189, "1_000": 4, "1d": [0, 98, 102, 105, 263, 288], "1e": [0, 4, 6, 16, 172, 328, 344, 348, 350, 351, 382, 422, 424, 434, 454, 455, 456, 457, 458, 459, 460, 463, 466, 468, 469, 470, 471, 472], "1e3": 479, "1st": 237, "2": [0, 1, 2, 4, 5, 6, 38, 99, 103, 117, 118, 134, 148, 151, 153, 154, 155, 156, 157, 158, 159, 168, 178, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 198, 204, 237, 245, 250, 291, 294, 296, 297, 298, 309, 313, 317, 324, 326, 327, 331, 334, 341, 351, 355, 356, 382, 391, 400, 401, 402, 403, 404, 405, 406, 407, 408, 412, 423, 424, 426, 433, 434, 449, 452, 454, 455, 457, 458, 459, 463, 466, 479, 480, 481, 482, 483, 484, 485, 486, 487], "20": [168, 189], "200": [5, 470], "2002": 5, "2011": 457, "2012": [455, 466], "2015": [337, 458, 460], "2019": [5, 459], "2020": 5, "2021": 5, "20397": 421, "20_000": 5, "21": [5, 472], "2104": 5, "223144": 421, "223404": 403, "225": 189, "225763": 430, "2302": 461, "23607": [189, 190], "24": 8, "24264": 189, "247": 5, "25": [379, 400], "25211": 404, "256": [1, 2, 6, 143], "256995": 430, "28": 168, "2d": [0, 99, 103, 118, 237, 328, 337], "2nd": 237, "2x": 484, "3": [0, 1, 2, 5, 8, 100, 104, 159, 178, 185, 186, 187, 189, 190, 309, 313, 332, 335, 341, 400, 403, 405, 412, 417, 456, 461, 476, 479, 482, 484, 485], "30": 456, "3118": 484, "32": [1, 5, 6, 90, 237, 238, 317, 327, 356, 382, 479], "330": 5, "33333": 400, "348587": 423, "363207": 402, "36788": 479, "379159": 403, "380709": 407, "39": 5, "390": 481, "3d": [0, 2, 100, 104, 328, 338, 400], "3f": [2, 6, 479], "3x": 2, "4": [0, 1, 2, 5, 116, 141, 143, 159, 164, 189, 237, 238, 265, 307, 313, 317, 326, 327, 328, 348, 355, 356, 380, 381, 399, 400, 402, 403, 404, 421, 479, 480, 482, 485, 487], "4096": [479, 481, 487], "40x": 1, "41421": 189, "417497": 408, "42": 314, "437": 5, "44": 5, "447214": 190, "458835": 404, "475": 5, "48095": 402, "4d": [1, 400], "4m": 1, "5": [0, 1, 2, 4, 5, 8, 189, 217, 240, 313, 326, 328, 336, 337, 338, 341, 345, 348, 355, 395, 400, 401, 404, 405, 412, 415, 433, 445, 449, 454, 466, 468, 469, 479, 481, 482], "50": [0, 193], "500": [5, 487], "5000": 2, "510826": 421, "512": [2, 3, 5, 399, 487], "534422": 407, "539245": 421, "53947": 402, "55": 1, "5701": 455, "573409": 430, "57771": 190, "579": 5, "5f": 4, "6": [1, 2, 5, 189, 265, 385, 399, 403, 412, 413, 417, 424, 434, 438, 466, 479, 482, 485], "61278": 402, "617261": 408, "628": 5, "633": 5, "64": [0, 1, 90, 116, 141, 164, 237, 238, 307, 317, 380, 381], "64331": 405, "666329": 405, "66667": 400, "67326": 439, "676": 1, "690": 5, "6967": 404, "7": [2, 5, 189, 237, 482], "702": [341, 413], "707107": 186, "71828": 479, "74166": 189, "74597": 189, "75": 400, "75596": 430, "75787": 404, "765166": 430, "773433": 430, "776856": 403, "793615": 405, "79854": 405, "7b": 5, "7m": 1, "8": [0, 1, 2, 5, 8, 189, 237, 317, 327, 348, 356, 399, 422, 455, 456, 457, 458, 459, 460, 466, 479, 482, 485, 487], "8192": [5, 168], "84804": 189, "863726": 408, "883935": 408, "890597": 403, "894427": 190, "89613": 402, "8gb": 5, "8x": 1, "9": [8, 189, 423, 455, 458, 459, 460, 461, 463, 469, 472, 484], "90041": 403, "912766": 403, "916291": 421, "95": 6, "982273": 407, "99": [461, 466], "995016": 402, "999": [458, 459, 460], "A": [0, 2, 5, 7, 8, 9, 68, 82, 94, 142, 143, 144, 146, 165, 178, 179, 184, 186, 187, 189, 190, 191, 194, 203, 204, 205, 210, 221, 237, 240, 241, 242, 244, 245, 246, 247, 248, 251, 252, 275, 279, 282, 299, 302, 303, 307, 308, 309, 310, 311, 312, 313, 314, 315, 324, 328, 337, 343, 344, 348, 350, 362, 366, 367, 370, 376, 377, 382, 388, 391, 399, 402, 403, 405, 413, 434, 435, 452, 454, 458, 460, 462, 463, 465, 470, 479, 480, 481, 483, 484], "AS": 163, "And": [5, 400], "As": [6, 38, 287, 324], "At": 93, "But": 487, "By": [5, 307, 370, 421, 481, 484], "For": [0, 1, 2, 5, 8, 38, 146, 163, 178, 189, 237, 314, 324, 328, 337, 341, 358, 363, 372, 375, 381, 386, 391, 400, 402, 403, 404, 405, 421, 449, 454, 476, 479, 480, 481, 482, 483, 484, 485, 486, 487], "If": [0, 1, 2, 5, 8, 15, 16, 17, 18, 26, 27, 28, 29, 78, 82, 83, 93, 95, 105, 108, 109, 110, 111, 117, 118, 121, 122, 123, 125, 126, 127, 136, 142, 145, 156, 157, 158, 161, 162, 165, 172, 183, 184, 185, 189, 194, 203, 204, 205, 207, 208, 216, 217, 221, 225, 229, 232, 233, 235, 236, 241, 245, 247, 256, 259, 273, 274, 275, 280, 284, 286, 287, 288, 291, 293, 294, 299, 300, 303, 305, 307, 311, 313, 328, 330, 331, 332, 333, 334, 335, 344, 350, 352, 363, 365, 375, 381, 383, 386, 388, 391, 400, 421, 423, 434, 456, 479, 480, 481, 483, 486, 487, 488], "In": [0, 1, 2, 5, 6, 38, 204, 237, 311, 324, 337, 344, 452, 455, 457, 458, 460, 461, 462, 478, 479, 480, 481, 483, 486, 487], "It": [2, 5, 8, 126, 165, 268, 299, 309, 313, 324, 377, 381, 462, 474, 484, 486], "Its": 324, "No": [2, 5, 186, 187], "Not": [94, 228, 479], "ON": [3, 8], "Of": 481, "On": [1, 479, 481, 483], "One": [147, 150, 156, 232, 261, 479, 481], "THE": 8, "That": 5, "The": [0, 1, 2, 3, 5, 6, 7, 8, 12, 13, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 37, 47, 51, 61, 62, 68, 78, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 115, 116, 117, 118, 121, 122, 123, 125, 126, 127, 128, 129, 130, 131, 133, 134, 135, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 179, 180, 181, 182, 185, 186, 187, 189, 190, 191, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 212, 213, 216, 217, 218, 219, 221, 222, 223, 224, 226, 228, 229, 230, 231, 232, 233, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 263, 264, 269, 270, 271, 272, 273, 274, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 307, 308, 309, 310, 311, 312, 313, 314, 317, 319, 326, 327, 328, 330, 331, 332, 333, 334, 335, 336, 337, 338, 340, 342, 343, 344, 348, 349, 350, 352, 355, 356, 358, 359, 363, 365, 369, 370, 371, 372, 375, 376, 377, 378, 380, 381, 382, 383, 386, 388, 391, 397, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 414, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 446, 449, 452, 454, 455, 456, 457, 458, 459, 460, 461, 464, 466, 467, 468, 471, 474, 479, 480, 481, 482, 483, 484, 485, 486, 487, 488], "Then": [4, 8], "There": [1, 2, 324, 400, 479], "These": [1, 2, 94, 236, 288, 423, 487], "To": [0, 2, 3, 4, 5, 6, 8, 216, 324, 449, 454, 479, 480, 481, 485], "With": 2, "_": [1, 3, 4, 5, 312, 324, 468, 469, 470, 471, 472, 476, 479, 483, 487], "__call__": [1, 5, 6, 324, 452], "__init__": [2, 5, 6, 9, 10, 11, 30, 112, 120, 315, 324, 452], "__main__": [2, 5], "__name__": [2, 5], "_a": 2, "_ext": 2, "_f": 189, "_in": [402, 403], "_out": [402, 403], "_p": 434, "_size": [326, 327, 355, 356], "_val": 416, "a1": 163, "a2": 163, "a_": 189, "a_max": [0, 93], "a_min": [0, 93], "a_ndim": 1, "a_shap": 1, "a_strid": 1, "a_view": 484, "ab": [0, 16, 172, 189, 299, 344, 348, 350, 357, 382, 413, 435, 479], "abil": 480, "abl": [2, 237], "about": [1, 2, 5, 6, 131, 210, 483, 487], "abov": [1, 2, 5, 237, 297, 324, 400, 459, 480, 481, 482, 483, 487], "absolut": [0, 12, 16, 172, 412, 413, 433], "acc": 313, "acceler": [2, 328], "access": [0, 5, 50, 324, 452, 463, 480, 483, 487], "accord": [0, 242, 304, 307, 378, 402, 403, 404, 405], "accordingli": 2, "accross": 8, "accumul": [313, 382], "accuraci": 6, "accustom": 5, "achiev": [324, 480], "across": [1, 2, 344, 480], "act": [2, 429], "action": 324, "activ": [2, 8, 211, 337, 397, 399, 415, 435, 445, 446, 448, 479], "actual": [5, 18, 365, 452, 483], "ad": [0, 1, 2, 4, 8, 142, 348, 452, 455, 456, 457, 458, 459, 460, 466, 480, 483, 486], "adadelta": 454, "adafactor": 454, "adagrad": 454, "adam": [454, 460, 461, 470, 471], "adamax": 454, "adamw": [454, 461], "adapt": [455, 456, 457, 480], "add": [0, 1, 2, 3, 5, 14, 38, 138, 199, 232, 237, 330, 331, 332, 333, 334, 335, 481, 487], "add_argu": 5, "add_depend": 2, "add_librari": 2, "addit": [0, 2, 5, 8, 13, 14, 142, 144, 146, 194, 328, 344, 350, 378, 382, 452, 481], "addmm": 0, "address": 2, "adjac": 337, "advanc": [5, 479], "advantag": 487, "advis": 484, "affin": [328, 344, 348, 350, 352, 381], "after": [2, 5, 6, 28, 159, 161, 164, 209, 233, 237, 328, 344, 350, 358, 359, 363, 365, 372, 375, 376, 377, 378, 399, 433, 479, 487], "after_1": 232, "after_2": 232, "after_i": 232, "after_n": 232, "afternoon": 5, "again": [5, 8, 324, 479], "against": 0, "aggreg": 378, "ago": 5, "ai": 112, "ainv": [188, 192], "albeit": 487, "algebra": 7, "algorithm": [400, 461], "alia": [96, 97, 341], "alibi": 324, "align": [184, 237, 327, 343, 349, 356], "align_corn": 400, "all": [0, 1, 2, 3, 6, 8, 16, 28, 38, 84, 85, 86, 94, 99, 100, 101, 103, 104, 112, 121, 122, 123, 140, 149, 152, 155, 158, 163, 164, 191, 204, 232, 233, 259, 278, 307, 324, 358, 359, 363, 366, 367, 368, 373, 375, 378, 391, 399, 400, 449, 452, 474, 476, 479, 482, 483, 485, 488], "all_avg": 480, "all_reduce_grad": 480, "all_sum": 480, "allclos": [0, 1, 143], "alloc": [2, 212, 216, 217, 452], "allow": [0, 1, 2, 178, 309, 324, 377, 452, 474, 480, 482, 485], "almost": 5, "alon": [2, 484], "along": [0, 2, 26, 27, 94, 95, 108, 109, 110, 111, 121, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 163, 164, 168, 185, 189, 236, 247, 256, 259, 273, 275, 279, 287, 288, 291, 292, 293, 294, 301, 324, 342, 383, 414], "alpha": [0, 2, 14, 237, 329, 339, 409, 410, 434, 436, 439, 459, 466], "alpha_": 2, "alreadi": [2, 3, 5, 480], "also": [0, 1, 2, 5, 6, 7, 8, 11, 13, 87, 88, 89, 119, 128, 129, 133, 149, 152, 155, 158, 166, 167, 180, 181, 182, 199, 206, 222, 224, 228, 234, 237, 255, 258, 283, 307, 308, 319, 324, 362, 376, 378, 380, 381, 389, 411, 439, 441, 448, 454, 479, 480, 481, 482, 483, 484, 485, 488], "altern": 476, "alwai": [1, 83, 211, 310, 481], "am": 5, "among": 2, "amount": [5, 213, 326, 355], "amus": 5, "an": [0, 1, 2, 3, 5, 6, 8, 10, 15, 17, 30, 84, 85, 86, 91, 98, 99, 100, 101, 102, 103, 104, 120, 125, 126, 127, 136, 140, 142, 146, 159, 162, 169, 173, 183, 189, 194, 217, 218, 223, 229, 230, 232, 235, 236, 237, 238, 247, 256, 257, 259, 260, 275, 278, 285, 287, 288, 291, 292, 296, 303, 305, 306, 310, 311, 312, 313, 324, 326, 327, 336, 341, 344, 349, 350, 352, 355, 356, 358, 378, 379, 381, 383, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 412, 436, 449, 454, 455, 465, 469, 474, 476, 478, 479, 480, 481, 482, 483, 484, 485, 486, 487, 488], "anaconda": 480, "anchor": 434, "angl": [115, 239, 351], "angular": [145, 386], "ani": [0, 1, 2, 5, 7, 18, 94, 310, 311, 312, 313, 314, 324, 341, 358, 359, 362, 371, 381, 399, 400, 449, 471, 478, 479, 481, 483, 485, 486, 487], "anonym": 479, "anoth": [0, 93, 178, 204, 283, 304, 317, 324, 358, 479, 481, 482, 487], "anwywher": 8, "anyhow": 5, "anymor": 5, "anyth": [5, 299, 483], "anytim": 483, "api": [1, 2, 341, 480, 481], "app": 8, "append": [5, 204, 479, 483], "appl": [2, 5, 7, 8, 487], "appli": [0, 38, 145, 146, 163, 191, 311, 312, 313, 324, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 337, 338, 339, 341, 342, 344, 345, 346, 347, 348, 350, 351, 352, 353, 354, 355, 356, 357, 359, 372, 379, 381, 382, 383, 384, 385, 387, 389, 390, 392, 393, 394, 395, 396, 397, 398, 400, 409, 410, 411, 414, 415, 416, 417, 418, 419, 420, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445, 446, 447, 449, 462, 465, 471, 474, 479, 480], "applic": [3, 8], "apply_fn": 359, "apply_gradi": 454, "apply_to_modul": [324, 363], "approach": [429, 481], "appropri": [2, 479], "approx": 341, "approxim": [16, 341, 411, 412, 413], "ar": [0, 1, 2, 4, 5, 6, 7, 8, 16, 18, 82, 90, 91, 93, 94, 101, 105, 112, 118, 125, 126, 136, 140, 143, 148, 149, 151, 152, 154, 155, 157, 158, 159, 164, 165, 172, 173, 174, 175, 176, 177, 178, 179, 186, 187, 189, 190, 194, 204, 217, 231, 232, 233, 237, 238, 240, 241, 242, 247, 248, 251, 252, 259, 265, 266, 278, 279, 287, 299, 302, 303, 307, 310, 311, 317, 328, 330, 331, 332, 333, 334, 335, 336, 337, 338, 344, 348, 350, 352, 365, 378, 381, 400, 421, 423, 424, 448, 452, 454, 461, 463, 478, 479, 480, 481, 482, 483, 484, 485, 486, 487], "arang": [0, 1, 189, 247, 317, 400, 482, 484], "arbitrari": [310, 452], "arbitrarili": [1, 94, 324, 478, 481, 485], "arc": 0, "arcco": 0, "arccosh": 0, "architectur": [5, 8, 210, 324, 377, 487], "archiv": 486, "arcsin": 0, "arcsinh": 0, "arctan": 0, "arctan2": 0, "arctanh": 0, "arg": [2, 5, 10, 18, 120, 136, 265, 266], "arg1": 178, "arg2": 178, "argmax": [0, 6], "argmin": 0, "argnam": [165, 299], "argnum": [2, 165, 299, 481], "argpars": 5, "argpartit": 0, "argsort": 0, "argument": [1, 31, 65, 79, 94, 136, 165, 299, 311, 312, 313, 324, 400, 476, 480, 481, 486, 487, 488], "argumentpars": 5, "ari": [84, 85, 86], "aris": 484, "arm": 8, "arm64": 8, "around": 5, "arr": [0, 262, 482], "arr_0": 486, "arrai": [0, 1, 2, 5, 6, 7, 10, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 115, 116, 117, 118, 121, 122, 125, 126, 127, 128, 129, 130, 131, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 221, 222, 223, 224, 225, 226, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 283, 284, 285, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 309, 324, 328, 349, 358, 365, 368, 373, 379, 400, 401, 402, 403, 404, 405, 406, 407, 408, 414, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 446, 449, 452, 455, 456, 457, 458, 459, 460, 461, 466, 467, 468, 469, 470, 471, 472, 479, 480, 481, 483, 484, 485, 487], "array_equ": [0, 16, 172], "arrayfir": 7, "arxiv": [5, 344, 348, 350, 357, 382, 413, 435, 455, 461], "as_strid": 0, "ascend": [186, 187], "ask": [5, 480], "assert": [1, 2, 143], "assign": [0, 2, 38, 452], "associ": [2, 265, 266, 483], "assum": [0, 2, 5, 90, 185, 186, 187, 190, 311, 324, 326, 327, 344, 355, 356], "astyp": [0, 1, 2, 5, 143, 358, 484], "atleast": 0, "atleast_1d": 0, "atleast_2d": 0, "atleast_3d": 0, "atol": [0, 16, 172], "atom": [1, 143], "atomic_fetch_add_explicit": 1, "atomic_output": [1, 143], "attach": 2, "attempt": 94, "attend": 378, "attent": [146, 363, 378, 391, 399], "attention_norm": 5, "attribut": [1, 9, 10, 11, 30, 315, 371, 452, 474], "audio": 400, "auto": [0, 2, 8], "autom": 481, "automat": [1, 2, 7, 143, 194, 480, 485, 486, 487], "autoregress": 5, "avail": [2, 4, 5, 6, 8, 10, 124, 214, 319, 487], "averag": [326, 327, 455, 456, 458, 459, 460, 480], "avgpool1d": 324, "avgpool2d": 324, "avoid": [1, 2, 370, 479], "awai": [2, 5], "awar": [479, 483], "ax": [0, 2, 15, 17, 26, 27, 79, 112, 138, 148, 149, 151, 152, 154, 155, 157, 158, 159, 171, 189, 203, 205, 207, 221, 232, 235, 259, 273, 278, 280, 284, 285, 291, 295, 300, 481], "axes_a": 0, "axes_b": 0, "axi": [0, 2, 5, 6, 15, 17, 26, 27, 28, 29, 33, 34, 35, 36, 41, 42, 43, 44, 56, 57, 58, 59, 63, 71, 74, 75, 76, 80, 95, 108, 109, 110, 111, 118, 121, 138, 142, 144, 147, 150, 153, 154, 155, 156, 157, 158, 159, 168, 185, 189, 203, 205, 207, 221, 223, 232, 233, 235, 236, 241, 247, 256, 259, 273, 274, 275, 278, 279, 280, 284, 285, 287, 288, 292, 293, 294, 295, 300, 301, 303, 326, 327, 342, 355, 356, 383, 414, 420, 422, 423, 427, 432, 434, 442, 443, 482], "axis1": [0, 46, 77, 118, 285, 294], "axis2": [0, 46, 77, 118, 285, 294], "axpbi": 2, "axpby_": 2, "axpby_gener": 2, "axpby_general_": 2, "axpby_impl": 2, "axpby_impl_acceler": 2, "b": [0, 1, 2, 3, 5, 13, 14, 16, 24, 82, 87, 88, 89, 90, 128, 129, 133, 143, 161, 163, 166, 167, 171, 172, 180, 181, 182, 185, 189, 199, 200, 202, 204, 206, 222, 224, 228, 231, 234, 237, 244, 255, 258, 283, 291, 299, 312, 313, 342, 352, 383, 400, 414, 481, 482, 483, 484, 485, 486, 487], "b1": 163, "b2": 163, "b_": [343, 349], "b_stride": 1, "ba": [458, 460], "back": [5, 112, 214, 484], "backend": [1, 8, 123, 124], "backward": [1, 479, 481], "bad": 483, "balanc": 429, "baltimor": 189, "bandwidth": [479, 480], "bar": 480, "base": [0, 2, 145, 189, 196, 198, 234, 386, 399, 452, 454, 460, 474, 476, 479, 482], "base_idx": 1, "basi": 474, "basic": [4, 260, 481], "batch": [5, 14, 90, 163, 164, 204, 245, 328, 330, 331, 332, 333, 334, 335, 337, 338, 343, 349, 378, 383, 400, 483], "batch_idx": 1, "batch_iter": [6, 454], "batch_siz": [6, 454], "batchnorm": 324, "becaus": [5, 211, 324, 483], "been": [0, 2, 5, 212, 483], "befor": [1, 2, 5, 8, 28, 143, 233, 362, 399, 463, 480, 482, 483], "before_1": 232, "before_2": 232, "before_i": 232, "before_n": 232, "beforehand": 231, "beggin": 259, "begin": [83, 184, 213, 237, 327, 343, 349, 356, 397, 415, 426, 433, 439, 445, 446], "behav": 112, "behavior": [245, 429, 482, 483], "behaviour": [112, 183, 184], "behind": 481, "being": [281, 324], "bell": 2, "below": [2, 8, 189, 296, 298, 317, 400, 483], "bench": 2, "benchmark": [2, 479], "benefici": [337, 338, 483], "best": 480, "beta": [0, 2, 14, 116, 141, 237, 328, 344, 348, 350, 433, 454, 458, 459, 460, 461], "beta_": 2, "beta_1": [456, 458, 459, 460, 461], "beta_2": [458, 459, 460, 461], "better": [481, 487], "between": [0, 2, 7, 93, 159, 399, 422, 425, 426, 429, 470, 480, 483, 484, 487], "beyond": [259, 468, 471], "bfloat16": [2, 11, 168, 317, 484], "bfloat16_t": 2, "bia": [5, 116, 141, 142, 164, 237, 238, 311, 324, 330, 331, 332, 333, 334, 335, 343, 349, 350, 352, 363, 365, 375, 378, 381, 383, 458, 459, 460, 463, 481], "bias": [0, 116, 141, 164, 237, 238, 343, 349, 363, 375, 378], "bicub": 400, "big": [1, 479], "bigger": [5, 456], "bilinear": [1, 400], "binari": [194, 262, 263, 264, 265, 266, 301, 397, 421, 446, 479], "binary_cross_entropi": [324, 479], "bit": [0, 116, 141, 164, 180, 237, 238, 258, 307, 317, 358, 380, 381, 382], "bitwis": [0, 87, 88, 89, 180, 258], "bitwise_and": 0, "bitwise_or": 0, "bitwise_xor": 0, "block": [0, 2, 5, 90, 399], "block_masked_mm": 0, "block_siz": [0, 90], "bn": 328, "bodi": [1, 143], "bool": [0, 1, 2, 15, 16, 17, 26, 27, 33, 34, 35, 36, 41, 42, 43, 44, 56, 57, 58, 59, 63, 75, 76, 78, 80, 82, 94, 101, 108, 109, 110, 111, 123, 124, 143, 145, 164, 172, 178, 183, 184, 189, 192, 194, 203, 205, 207, 208, 214, 217, 221, 235, 238, 280, 284, 300, 328, 330, 331, 332, 333, 334, 335, 343, 344, 348, 349, 350, 352, 358, 362, 363, 365, 370, 372, 375, 378, 381, 383, 386, 391, 399, 400, 421, 424, 456, 467], "bool_": [11, 317], "boolean": [0, 16, 82, 172, 173, 174, 175, 176, 177, 178, 200, 201, 202, 317, 374, 482], "both": [1, 2, 13, 87, 88, 89, 128, 129, 133, 166, 167, 178, 180, 181, 182, 189, 199, 206, 222, 224, 228, 234, 241, 255, 258, 283, 307, 326, 327, 348, 349, 355, 356, 454, 479, 480, 481, 485, 487], "bottom": 400, "bound": [0, 248, 251, 252, 341, 408, 479, 482, 487], "boundari": 470, "bracket": 5, "brain": 317, "break": 484, "bregler": 337, "broadcast": [0, 2, 13, 16, 87, 88, 89, 91, 93, 128, 129, 133, 162, 166, 167, 172, 180, 181, 182, 199, 204, 206, 222, 224, 228, 234, 236, 240, 241, 245, 251, 252, 255, 258, 283, 288, 304, 378], "broadcast_arrai": [0, 2], "broadcast_to": 0, "broadcasted_input": 2, "brought": 7, "btl_tcp_link": 480, "buffer": [1, 2, 211, 484], "bui": 5, "build": [3, 5, 7, 404, 452, 479], "build_ext": [2, 8], "build_shared_lib": [2, 8], "built": [1, 2, 8, 483], "bundl": 5, "byte": [51, 61, 211, 212, 213, 216, 217, 218, 317], "c": [0, 1, 2, 5, 14, 189, 326, 327, 328, 330, 331, 332, 333, 334, 335, 337, 338, 348, 349, 355, 356, 484, 485, 487], "c_": [349, 461], "c_in": [98, 99, 100, 101, 102, 103, 104], "c_j": [326, 327, 355, 356], "c_out": [98, 99, 100, 101, 102, 103, 104], "c_pad": 1, "c_t": [349, 461], "cach": [5, 8, 209, 211, 212, 216, 479], "calcul": [189, 421, 424, 430, 456], "call": [2, 3, 5, 6, 31, 126, 161, 209, 213, 324, 340, 363, 375, 380, 388, 452, 454, 463, 479, 480, 481, 483], "callabl": [94, 112, 143, 165, 179, 299, 302, 303, 307, 308, 310, 311, 312, 313, 358, 359, 362, 370, 383, 388, 399, 401, 402, 403, 404, 405, 406, 407, 408, 455, 456, 457, 458, 459, 460, 461, 466, 467, 468, 469, 470, 471, 472], "can": [1, 2, 3, 5, 7, 8, 13, 18, 65, 79, 83, 87, 88, 89, 94, 118, 119, 120, 128, 129, 133, 136, 166, 167, 180, 181, 182, 189, 199, 206, 218, 222, 224, 228, 234, 240, 241, 248, 251, 252, 255, 258, 263, 283, 294, 299, 313, 324, 327, 340, 341, 356, 362, 375, 380, 388, 400, 423, 449, 452, 454, 462, 463, 476, 478, 479, 480, 481, 482, 483, 484, 485, 486, 487, 488], "cannot": [5, 93, 482, 484], "captur": [2, 3, 94, 219, 220, 324, 479], "care": [5, 483], "carri": 2, "cartesian": 208, "case": [2, 5, 121, 122, 123, 125, 126, 127, 149, 152, 153, 155, 156, 157, 158, 159, 183, 184, 185, 186, 187, 188, 190, 191, 192, 204, 257, 278, 327, 337, 356, 397, 415, 433, 439, 445, 446, 462, 463, 479, 481, 485, 486, 487, 488], "cast": [2, 37, 156, 157, 158, 194, 358, 370, 484], "caster": 2, "categor": 5, "categori": [11, 178, 317], "catlas_saxpbi": 2, "caus": [324, 479, 483], "causal": 5, "caution": 83, "cd": [3, 8], "cdf": [242, 341, 411], "cdot": [413, 422, 425, 441], "ceil": 0, "ceildiv": 1, "cell": 349, "celu": 324, "certain": [2, 372, 479], "chang": [83, 94, 268, 301, 376, 381, 400, 426, 433, 479, 484], "channel": [1, 98, 99, 100, 101, 102, 103, 104, 328, 330, 331, 332, 333, 334, 335, 337, 338], "channel_idx": 1, "charact": 310, "check": [0, 2, 8, 82, 124, 178, 186, 187, 214, 365, 481, 482], "checklist": 480, "checkout": [3, 479], "checkpoint": [399, 454], "chen": 461, "child": 377, "children": 324, "chip": 8, "choleski": 184, "choos": [5, 145, 386], "chosen": 131, "clamp": 159, "clang": 8, "clariti": 481, "class": [2, 5, 6, 9, 10, 11, 30, 112, 120, 315, 325, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350, 351, 352, 353, 354, 355, 356, 357, 378, 379, 380, 381, 382, 383, 384, 385, 386, 387, 388, 389, 390, 391, 392, 393, 394, 395, 396, 397, 398, 399, 400, 409, 410, 411, 412, 413, 414, 415, 416, 417, 418, 419, 420, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445, 446, 447, 452, 455, 456, 457, 458, 459, 460, 461, 466, 467, 474], "class_pred": 307, "classif": [404, 405], "classifi": 6, "classmethod": [380, 381], "clear": 209, "click": 8, "clip": [0, 309, 421, 456], "clip_threshold": 456, "clipped_grad": 309, "clone": 8, "close": [4, 7, 8, 16, 172], "closer": 311, "cmake": [3, 8], "cmake_arg": 3, "cmake_build_parallel_level": 8, "cmake_build_typ": 8, "cmake_current_list_dir": 2, "cmake_host_system_processor": 8, "cmake_library_output_directori": 2, "cmakebuild": 2, "cmakeextens": 2, "cmakelist": 2, "cmdclass": 2, "co": [0, 2, 112, 391, 481], "code": [1, 143, 479, 480, 483], "coeffici": [2, 455, 456, 458, 459, 460, 461], "col": 296, "col_contigu": 2, "cold": 8, "collect": [2, 311, 312, 478], "column": [2, 140, 169, 186, 237], "com": [8, 480], "combin": [5, 191, 313], "come": [2, 5, 480, 481], "command": [2, 3, 8, 480], "command_buff": 2, "common": [2, 454, 479, 483], "commonli": [6, 376, 449, 479], "commun": [7, 120, 123, 124], "compar": [2, 82, 479], "comparison": [16, 133, 166, 167, 181, 182, 228], "compat": [5, 241, 245, 341, 486], "compil": [0, 3, 7, 8, 119, 132, 143, 480, 481, 483], "compiled_fun": 479, "compiled_grad_fn": 479, "complet": [4, 5, 8, 217, 376, 377, 481, 487], "complex": [2, 96, 97, 154, 155, 156, 157, 158, 170, 186, 187, 253, 310, 317, 324, 377, 479, 481], "complex64": [2, 11, 317], "complex64_t": 2, "complexflo": 11, "compon": [2, 5], "compos": [7, 324, 479, 481, 485], "composit": 485, "compress": 266, "compromis": 5, "comput": [0, 1, 2, 4, 5, 6, 7, 8, 108, 109, 110, 111, 112, 116, 131, 139, 141, 145, 165, 179, 183, 184, 185, 186, 187, 188, 189, 192, 199, 207, 231, 237, 255, 273, 280, 281, 291, 299, 300, 302, 308, 324, 328, 343, 344, 348, 349, 350, 363, 376, 381, 382, 386, 399, 402, 403, 404, 405, 412, 413, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 454, 455, 456, 458, 459, 460, 461, 465, 479, 480, 481, 485, 487], "computation": 483, "compute_encod": 2, "concaten": [0, 5, 121], "concept": 452, "concis": 5, "concret": [2, 343, 349, 352, 383, 483, 487], "conda": [8, 480], "condit": [0, 2, 304, 487], "config": [2, 480], "configu": 454, "configur": [116, 141, 480], "confirm": 480, "confus": 6, "conj": 97, "conjug": [0, 96], "connect": 480, "consecut": [145, 237, 386], "consequ": 5, "consid": [5, 16, 82, 172, 310, 311, 312, 344, 478], "consider": 479, "const": [0, 1, 2, 424], "constant": [0, 2, 5, 8, 142, 144, 232, 324, 328, 344, 350, 382, 424, 434, 466, 468, 479, 484], "constant_valu": 232, "constitut": 311, "construct": [0, 2, 6, 45, 117, 162, 229, 292, 305], "consum": 483, "contain": [2, 5, 8, 28, 29, 68, 94, 118, 131, 153, 154, 155, 163, 164, 186, 189, 200, 201, 202, 237, 275, 304, 309, 324, 362, 364, 365, 371, 399, 430, 449, 452, 479, 480, 481], "content": [8, 362, 479], "context": 282, "contigu": [1, 2, 83, 143], "continu": [329, 409, 481], "contract": [0, 131], "contrast": 459, "contribut": 2, "contriv": [481, 487], "control": [0, 351, 476, 483], "conv": 105, "conv1d": [0, 324], "conv2d": [0, 324], "conv3d": [0, 324], "conv_gener": 0, "conv_transpose1d": 0, "conv_transpose2d": 0, "conv_transpose3d": 0, "conveni": [1, 2, 6, 178], "convent": [18, 105, 130, 131, 400, 459], "convers": 7, "convert": [0, 1, 2, 78, 84, 85, 86, 115, 159, 239, 380, 381, 483, 484, 485], "convolut": [0, 98, 99, 100, 101, 102, 103, 104, 105, 330, 331, 332, 333, 334, 335, 337, 338], "convolv": [98, 99, 100, 101, 102, 103, 104], "convtranspose1d": 324, "convtranspose2d": 324, "convtranspose3d": 324, "coordin": [0, 208], "copi": [0, 1, 2, 5, 7, 233, 274, 484], "copy_inplac": 2, "copytyp": 2, "core": [1, 2, 3, 4, 5, 6, 307, 324, 326, 327, 328, 348, 355, 356, 365, 368, 370, 373, 400, 401, 402, 403, 404, 405, 406, 407, 408, 421, 423, 430, 449, 452, 454, 479, 480, 484, 485], "corner": 400, "correct": [2, 8, 458, 459, 460, 482, 483], "correctli": 38, "correl": [101, 337], "correspond": [0, 1, 2, 15, 17, 78, 93, 116, 118, 141, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 186, 203, 205, 221, 235, 284, 291, 303, 311, 481], "cos_first": 391, "cosh": [0, 429], "cosin": [0, 19, 20, 106, 107, 422, 468, 470, 481], "cosine_decai": [454, 470], "cosine_similarity_loss": 324, "cost": [8, 456, 480, 483], "costli": 483, "cot": 1, "cot_index": 1, "cotan": [2, 112], "cotang": [1, 2, 112, 302], "could": [5, 324], "count": [324, 470], "counter": 476, "cours": 481, "coursera": 466, "cov": 245, "covari": [245, 328], "cover": 2, "cpp": 2, "cpu": [7, 8, 186, 187, 190, 487], "cpython": 2, "crash": [83, 479], "creat": [0, 2, 5, 8, 83, 123, 140, 169, 282, 324, 452, 454, 470, 479, 482, 484], "create_additive_causal_mask": 5, "criteria": 2, "cross": [6, 101, 421, 423], "cross_entropi": [6, 324], "crowd": 5, "cry": 5, "cubic": 400, "cummax": 0, "cummin": 0, "cumprod": 0, "cumsum": 0, "cumul": [0, 83, 108, 109, 110, 111], "current": [5, 7, 8, 83, 90, 100, 103, 104, 127, 210, 212, 237, 313, 324, 456, 480, 483], "custom": [7, 112, 143, 399], "custom_decod": 399, "custom_encod": 399, "custom_funct": 1, "custom_kernel_myexp_float": 1, "custom_tim": 2, "cvpr": 337, "cycl": 478, "d": [0, 1, 2, 5, 100, 104, 117, 118, 171, 189, 204, 208, 231, 287, 294, 296, 297, 298, 314, 332, 335, 338, 343, 349, 383, 455, 458, 460, 487], "d1": 487, "d2": 487, "d2fdx2": 481, "d_i": 352, "dampen": 467, "darwin": 2, "data": [0, 2, 6, 7, 10, 18, 125, 140, 156, 157, 162, 169, 193, 225, 229, 242, 251, 294, 296, 301, 305, 338, 401, 402, 403, 404, 405, 406, 407, 408, 479, 480, 482, 484], "dataset": [4, 480, 483], "datatyp": 51, "dbuild_shared_lib": 8, "dcmake_build_typ": 8, "ddof": [0, 75, 80, 280, 300], "deal": 479, "debug": [1, 3, 480], "debugg": 7, "decai": [456, 459, 461, 467, 468, 469, 472], "decay_r": [456, 469, 472], "decay_step": 468, "decent": 6, "decid": [311, 362], "decim": [0, 66, 260], "declar": 2, "decltyp": 1, "decod": 399, "decomposit": [183, 184, 191], "decor": [1, 112], "decoupl": 459, "deep": [328, 402, 403, 404, 405], "def": [1, 2, 4, 5, 6, 112, 143, 299, 324, 452, 479, 480, 481, 482, 483, 484, 487], "default": [1, 2, 8, 14, 15, 16, 17, 18, 26, 27, 28, 29, 82, 83, 90, 94, 95, 98, 99, 100, 101, 102, 103, 104, 112, 113, 114, 116, 117, 118, 121, 122, 123, 125, 126, 127, 140, 141, 143, 145, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 163, 164, 165, 168, 169, 172, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 203, 205, 207, 208, 216, 217, 218, 221, 225, 229, 232, 233, 235, 237, 238, 240, 241, 242, 244, 245, 246, 247, 248, 250, 251, 252, 256, 257, 260, 267, 268, 274, 275, 278, 279, 280, 282, 284, 286, 291, 293, 294, 295, 296, 297, 298, 299, 300, 303, 305, 307, 317, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 339, 342, 343, 345, 348, 349, 351, 352, 355, 356, 358, 363, 365, 370, 372, 375, 378, 379, 380, 381, 383, 386, 391, 395, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 414, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 452, 455, 456, 457, 458, 459, 460, 461, 466, 467, 468, 476, 478, 479, 481, 484, 486, 488], "default_devic": 488, "default_stream": 488, "defin": [1, 2, 4, 5, 6, 8, 112, 126, 143, 164, 185, 189, 238, 307, 310, 484], "definit": [112, 183, 184, 245], "degre": [0, 239, 434], "delta": [426, 455], "delv": [404, 405], "demonstr": 484, "denomin": [348, 422, 455, 457, 458, 459, 460, 466], "dens": [208, 487], "depend": [0, 2, 3, 4, 8, 78, 189, 343, 349, 383, 480, 482, 486, 487], "depth": [310, 332, 335, 338, 481], "dequant": [0, 237], "deriv": [2, 481, 483], "descend": 360, "descent": [467, 479, 483], "describ": [2, 483], "descript": [2, 5, 317], "design": [1, 4, 7, 476, 487], "destin": [0, 2, 60, 127, 223, 236], "destroi": 479, "detach": 481, "detail": [1, 2, 10, 216, 324, 337, 386, 391, 400, 402, 403, 404, 405, 455, 457, 458, 460, 461, 482, 485], "determin": [0, 2, 118, 245, 313, 317, 369, 486], "dev": [2, 8], "develop": [2, 8], "developer_dir": 8, "deviat": [0, 246, 280, 402, 404, 407], "deviatoin": 0, "devic": [1, 2, 7, 8, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 113, 114, 115, 116, 117, 118, 121, 122, 125, 126, 127, 128, 129, 130, 133, 134, 135, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 210, 217, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 267, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 300, 301, 304, 305, 306, 315, 487, 488], "device_info": 218, "devicetyp": 9, "df": 484, "dfdx": [481, 482], "dft": [147, 148, 149, 150, 151, 152, 156, 157, 158], "dhwc": 338, "diag": [0, 191], "diagon": [0, 45, 117, 140, 294, 296, 297, 298], "dict": [94, 136, 194, 210, 263, 264, 265, 309, 368, 373, 376, 377, 452, 454, 462, 463, 465, 478, 481, 486], "dict_kei": [311, 463], "dictionari": [5, 94, 194, 210, 263, 264, 309, 310, 313, 324, 362, 371, 376, 377, 464, 478, 486], "did": 5, "diff": 2, "differ": [7, 178, 283, 301, 433, 481], "differenti": [1, 2, 7, 329, 409], "difficult": 481, "difficulti": [402, 403], "dilat": [0, 98, 99, 100, 101, 102, 103, 104, 330, 331, 333, 334], "dim": [1, 5, 145, 146, 340, 344, 348, 350, 378, 380, 382, 386, 391, 399], "dimens": [0, 1, 2, 5, 15, 17, 26, 27, 62, 68, 78, 84, 85, 86, 94, 99, 100, 101, 103, 104, 118, 138, 145, 154, 155, 157, 158, 159, 163, 164, 171, 183, 184, 186, 187, 188, 189, 190, 191, 192, 203, 204, 205, 207, 221, 235, 236, 237, 241, 250, 280, 284, 288, 291, 295, 300, 328, 330, 331, 332, 333, 334, 335, 337, 338, 342, 343, 344, 348, 349, 350, 378, 382, 383, 386, 399, 400, 414, 423, 479, 481], "dimension": [30, 142, 144, 147, 148, 149, 150, 151, 152, 156, 157, 158, 326, 327, 328, 330, 331, 332, 333, 334, 335, 340, 352, 355, 356, 380, 381, 391, 482, 484], "direct": [2, 5, 360, 461, 487], "directli": [2, 5, 83], "directori": [2, 5, 8], "disabl": [119, 216, 479], "disable_compil": 479, "disappoint": 5, "discard": [5, 310], "discov": 8, "discoveri": 461, "discret": [105, 147, 148, 149, 150, 151, 152, 156, 157, 158, 340, 380], "discuss": 2, "disk": 5, "dispatch": 2, "dispatchthread": [1, 2], "displai": 324, "distanc": [5, 434], "distribut": [7, 8, 240, 241, 242, 244, 245, 246, 251, 252, 352, 402, 403, 404, 405, 407, 408, 424, 427, 432, 434, 449], "diverg": 427, "divid": [0, 2, 38, 161, 237, 255, 480], "divis": [0, 128, 161, 237, 255], "divisor": [280, 300], "divmod": 0, "dloss_dw": 481, "dloss_dx": 481, "dlpack": 484, "dlvalu": 299, "dmlx_build_cpu": 8, "dmlx_build_gguf": 8, "dmlx_build_safetensor": 8, "dmlx_metal_debug": 3, "dmlx_metal_jit": 8, "do": [0, 2, 5, 8, 301, 324, 364, 375, 449, 452, 459, 479, 480, 481, 483], "doc": [2, 6, 480], "document": [2, 3, 65, 79, 143, 263, 264, 317, 479, 481, 482], "doe": [0, 2, 3, 5, 8, 211, 301, 309, 324, 479, 482, 483, 484], "doesn": [2, 324], "domain": [251, 480], "don": [1, 8, 479, 487], "done": [324, 336, 382, 479, 480, 483, 484], "dot": [183, 188, 192, 291, 310, 367, 378], "doubl": [0, 5], "doubt": 5, "down": [5, 309], "dparam": 299, "draw": 241, "drop": 362, "dropout": [324, 337, 338, 372, 399, 479], "dropout2d": 324, "dropout3d": 324, "dst": 127, "dt": 134, "dtype": [0, 1, 2, 5, 11, 18, 30, 37, 38, 78, 81, 125, 126, 140, 143, 159, 162, 169, 178, 186, 187, 189, 190, 193, 229, 242, 244, 245, 246, 248, 251, 252, 294, 296, 301, 305, 317, 370, 400, 401, 402, 403, 404, 405, 406, 407, 408, 421, 423, 430, 468, 469, 470, 471, 472, 479, 480, 481, 482, 484, 485, 486], "dtypecategori": [178, 317], "dual": 429, "duchi": 457, "dure": [3, 94, 336, 337, 338, 400, 484], "dx": 112, "dy": 112, "dyld": 480, "dyld_library_path": 480, "dylib": 2, "dynam": 483, "e": [2, 6, 8, 112, 134, 143, 163, 164, 179, 269, 328, 330, 331, 332, 333, 334, 335, 337, 338, 344, 348, 350, 363, 382, 419, 420, 442, 443, 448, 454, 457, 479, 483, 488], "e5": 317, "e8": 317, "each": [0, 1, 2, 68, 116, 136, 141, 145, 164, 178, 183, 184, 186, 187, 188, 191, 192, 204, 208, 232, 237, 238, 241, 256, 265, 266, 275, 292, 295, 301, 303, 304, 337, 338, 340, 343, 344, 349, 383, 386, 399, 421, 423, 476, 479, 480, 483], "eager": 483, "earli": 337, "earlier": 2, "eas": 5, "easi": [2, 324, 480], "easier": [1, 483], "edg": [93, 232, 400, 479], "edit": [8, 377], "effect": [337, 479, 483], "effici": [5, 7, 163, 337, 386, 483, 485], "eigenvalu": [186, 187], "eigenvector": 186, "einstein": [130, 131], "einsum": 131, "either": [8, 13, 65, 78, 79, 87, 88, 89, 93, 128, 129, 133, 161, 166, 167, 180, 181, 182, 189, 199, 204, 206, 222, 224, 228, 234, 255, 258, 283, 299, 327, 356, 388, 400, 404, 405, 484], "elem": [1, 143], "elem_to_loc": [1, 2], "element": [0, 1, 2, 12, 13, 19, 20, 21, 22, 23, 24, 25, 28, 70, 83, 87, 88, 89, 92, 106, 107, 108, 109, 110, 111, 116, 128, 129, 133, 134, 135, 137, 139, 140, 141, 160, 161, 164, 166, 167, 172, 173, 174, 175, 176, 177, 180, 181, 182, 195, 196, 197, 198, 199, 200, 201, 202, 206, 208, 222, 224, 226, 228, 233, 234, 237, 238, 254, 255, 256, 258, 259, 261, 269, 270, 271, 272, 276, 277, 283, 287, 289, 290, 293, 299, 301, 304, 329, 336, 337, 338, 343, 347, 349, 357, 379, 383, 386, 390, 409, 416, 417, 419, 420, 435, 436, 438, 441, 442, 443, 444, 479, 481], "elementwis": [1, 96, 97], "elif": 5, "ellipsi": 482, "elman": 383, "els": [0, 2, 5, 324, 363, 480, 483], "elsewher": [296, 482], "elu": [324, 439], "emb": [5, 340, 380, 391], "embed": [5, 307, 324, 380, 386, 391, 422], "empti": [127, 245], "enabl": [3, 5, 8, 94, 132, 467], "encod": [2, 145, 386, 391, 399, 423], "encount": [2, 481], "end": [118, 184, 214, 237, 259, 327, 343, 349, 356, 397, 415, 426, 433, 439, 445, 446, 468, 471], "end_axi": [0, 49, 159], "end_encod": 2, "endif": 2, "endswith": 363, "enhanc": [5, 386, 483], "enjoi": 2, "enough": [2, 483], "ensur": [0, 1, 2, 8, 143, 309, 429, 480], "ensure_row_contigu": [1, 143], "enter": 5, "entir": [15, 17, 26, 27, 203, 205, 207, 221, 235, 280, 284, 300, 337, 338], "entri": [0, 247, 337, 338], "entropi": [6, 421, 423], "enumer": 324, "environ": [8, 119, 132, 480], "ep": [4, 142, 144, 328, 344, 348, 350, 382, 422, 424, 434, 454, 455, 456, 457, 458, 459, 460, 466], "epoch": 6, "epsilon": [328, 344, 348, 350, 382, 422, 424, 455, 457, 458, 459, 460, 466], "epsilon_1": 456, "epsilon_2": 456, "equal": [0, 1, 16, 28, 82, 140, 167, 172, 182, 228, 233, 248, 275, 348, 352], "equal_nan": [0, 16, 82, 172], "equat": [130, 131], "equival": [0, 2, 31, 65, 79, 126, 129, 161, 164, 168, 287, 329, 339, 341, 345, 346, 347, 353, 354, 377, 379, 381, 384, 385, 387, 389, 392, 393, 394, 395, 396, 398], "erf": [0, 135, 479], "erfinv": 0, "error": [0, 2, 8, 123, 134, 135, 217, 218, 275, 341, 411, 412, 413, 429, 431, 481, 484], "error_norm": 4, "estim": [458, 460], "eta": 461, "etc": [2, 237, 324, 400, 480], "eval": [2, 3, 4, 5, 6, 324, 452, 454, 479, 480, 481, 483, 485], "eval_cpu": 2, "eval_fn": 6, "eval_gpu": 2, "evalu": [2, 5, 6, 7, 127, 136, 179, 302, 324, 361, 372, 452, 454, 479, 485], "even": [1, 2, 5, 94, 479, 483, 484], "evenli": [0, 193], "everi": [237, 311, 454, 472, 481], "everyth": [5, 480], "everywher": 0, "exact": [412, 413], "exactli": [2, 5, 145, 365, 481], "exampl": [0, 3, 4, 5, 6, 8, 18, 38, 112, 143, 159, 178, 186, 187, 189, 190, 282, 287, 309, 312, 313, 324, 326, 327, 328, 348, 355, 356, 363, 365, 372, 375, 400, 401, 402, 403, 404, 405, 406, 407, 408, 421, 423, 430, 449, 454, 463, 468, 469, 470, 471, 472, 476, 481, 482, 483, 484, 485, 486], "exce": 309, "exceed": 217, "except": [7, 140, 153, 154, 156, 157, 158, 344, 365, 482, 484], "exclud": [236, 288], "exclus": [0, 83, 89], "execut": [2, 8, 84, 85, 86, 213, 484, 487], "exist": [2, 3, 5, 363, 375], "exp": [0, 1, 139, 143, 199, 203, 242, 273, 329, 339, 390, 409, 410, 427, 439, 440, 444, 479, 487], "exp_elementwis": [1, 143], "expand_dim": 0, "expect": [2, 5, 330, 331, 332, 333, 334, 335, 336, 337, 338, 391, 399, 424, 479, 482], "expens": 399, "expensive_fun": 483, "experiment": 484, "explain": 2, "explicit": [2, 463, 476, 484], "explicitli": [163, 324, 476], "explor": 8, "expm1": 0, "exponenti": [0, 137, 139, 329, 339, 387, 409, 410, 439, 469], "exponential_decai": 454, "export": 8, "ext_modul": 2, "extend": [2, 232], "extens": [7, 194, 219, 369, 486], "extern": 484, "extra": [1, 311, 312], "extract": [0, 5, 45, 117, 118, 324, 362, 452, 480], "extras_requir": 2, "extrem": [482, 483], "ey": [0, 5, 188, 192], "f": [0, 2, 4, 6, 112, 189, 324, 349, 459, 479, 484], "f_jvp": 112, "f_t": 349, "f_vjp": 112, "f_vmap": 112, "face": 5, "factor": [2, 14, 168, 183, 184, 190, 400, 423, 469, 472], "fall": [2, 112], "fallback": 2, "fals": [0, 1, 2, 5, 15, 16, 17, 26, 27, 33, 34, 35, 36, 41, 42, 43, 44, 56, 57, 58, 59, 63, 75, 76, 80, 82, 94, 101, 108, 109, 110, 111, 123, 143, 172, 178, 183, 184, 189, 192, 194, 203, 205, 207, 208, 217, 221, 235, 280, 284, 300, 304, 307, 310, 311, 312, 313, 317, 344, 348, 350, 352, 363, 365, 375, 378, 381, 386, 391, 399, 400, 421, 424, 456, 467, 484], "famili": 5, "fan": [402, 403, 404, 405], "fan_in": [402, 403, 404, 405], "fan_out": [402, 403, 404, 405], "far": 454, "fast": [1, 7, 341, 413, 480, 487], "faster": [1, 2, 8, 129, 411, 421, 479, 481], "featur": [1, 7, 98, 99, 100, 101, 102, 103, 104, 145, 328, 343, 344, 348, 349, 350, 352, 381, 382, 383, 386, 399, 400, 479, 483], "feed": 5, "feed_forward": 5, "feedforward": [402, 403], "feel": 5, "fetch": 1, "few": [1, 2, 5, 6, 7, 8, 480, 483, 485], "ffn": 5, "ffn_norm": 5, "fft": 7, "figur": 480, "file": [5, 8, 194, 262, 263, 264, 265, 266, 365, 369, 480, 481, 486], "file_or_weight": 365, "fill": [0, 2, 162, 230, 296, 306, 401, 402, 403, 404, 405, 407, 408], "filter": [0, 105, 330, 331, 332, 333, 334, 335, 358, 362], "filter_and_map": 324, "filter_fn": [358, 362], "final": [2, 4, 5, 6, 168, 468, 471], "find": [2, 4, 8, 480], "find_packag": 2, "finder": 8, "fine": [476, 483], "finetun": 324, "finish": 2, "finit": [0, 173, 225], "first": [0, 1, 2, 3, 4, 5, 6, 8, 118, 121, 159, 165, 178, 180, 191, 200, 202, 204, 233, 250, 258, 285, 291, 294, 299, 310, 312, 313, 324, 327, 344, 356, 400, 422, 430, 456, 458, 459, 460, 463, 479, 481, 484, 487], "first_lay": 483, "fit": [2, 237, 487], "five": 479, "fix": [2, 5, 8, 483], "flag": [2, 8, 479, 484], "flat": [163, 164, 310, 314], "flat_param": 265, "flatten": [0, 28, 29, 108, 109, 110, 111, 189, 231, 233, 236, 256, 259, 274, 287, 288, 293, 310], "flexibl": 7, "flexibli": 377, "flip": [0, 101, 105], "float": [0, 1, 2, 11, 14, 16, 18, 78, 142, 143, 144, 145, 146, 161, 162, 168, 172, 178, 189, 225, 238, 240, 244, 246, 309, 317, 328, 336, 337, 338, 344, 348, 350, 358, 370, 382, 386, 391, 397, 399, 400, 401, 402, 403, 404, 405, 407, 408, 422, 423, 424, 426, 430, 433, 434, 445, 446, 455, 456, 457, 458, 459, 460, 461, 466, 467, 468, 469, 471, 472], "float16": [1, 2, 11, 143, 168, 194, 317, 358, 483, 484], "float16_t": [1, 2], "float32": [0, 1, 2, 11, 18, 140, 143, 146, 168, 169, 178, 186, 187, 189, 190, 193, 229, 242, 244, 245, 246, 251, 252, 296, 305, 317, 400, 401, 402, 403, 404, 405, 406, 407, 408, 421, 423, 430, 468, 469, 470, 471, 472, 479, 480, 481, 482, 483, 484, 485, 486], "float64": 178, "floor": [0, 1, 161], "floor_divid": 0, "flow": [0, 281, 483], "flush": 2, "fn": [308, 311, 312, 313, 485], "follow": [1, 2, 5, 6, 7, 8, 18, 105, 116, 141, 163, 189, 232, 237, 312, 324, 412, 413, 427, 455, 456, 457, 458, 459, 460, 461, 467, 476, 479, 480, 481, 487], "foo": 480, "food": 5, "forc": [5, 6, 324, 480, 485], "forg": 8, "formal": [116, 141, 237], "format": [5, 194, 262, 263, 264, 265, 266, 484], "formul": [329, 339], "formula": 433, "forth": 400, "forward": [1, 2, 299, 479, 483], "found": 362, "four": 328, "fourier": [147, 148, 149, 150, 151, 152, 156, 157, 158], "frac": [134, 237, 269, 326, 327, 328, 336, 337, 338, 344, 348, 350, 352, 355, 356, 382, 390, 402, 403, 404, 405, 422, 424, 426, 429, 440, 442, 443, 455, 457, 458, 459, 460, 466], "fraction": 18, "framework": [2, 7], "free": 216, "freez": [324, 375, 452], "freq": 145, "frequenc": [145, 386, 391], "frequent": [479, 483], "friend": 5, "fro": 189, "frobeniu": 189, "from": [0, 1, 2, 5, 6, 7, 83, 115, 116, 118, 121, 122, 125, 126, 127, 141, 143, 154, 155, 157, 158, 162, 163, 168, 189, 194, 204, 208, 213, 216, 230, 237, 239, 240, 241, 242, 243, 244, 248, 251, 265, 278, 281, 283, 287, 288, 293, 294, 304, 306, 310, 311, 312, 313, 314, 324, 352, 363, 365, 378, 402, 403, 404, 405, 407, 408, 424, 433, 449, 454, 478, 479, 480, 481, 483, 484, 485, 486, 487], "from_embed": 380, "from_linear": 381, "front": 2, "frozen": [324, 363, 373, 375, 381, 452], "fuction": 129, "full": [0, 1, 2, 6, 65, 79, 105, 143, 273, 376, 377, 424, 479, 480, 483], "full_turn": 391, "fulli": [2, 7, 480, 484, 487], "fun": [94, 165, 179, 299, 302, 303, 479, 482, 483, 487], "fun1": 483, "func": 383, "function": [0, 1, 2, 3, 4, 5, 6, 7, 16, 18, 83, 94, 112, 129, 134, 135, 143, 165, 172, 179, 183, 184, 186, 187, 188, 189, 190, 191, 192, 204, 218, 269, 299, 302, 303, 308, 309, 311, 312, 313, 324, 329, 339, 341, 342, 345, 346, 347, 353, 354, 357, 359, 363, 370, 375, 379, 383, 384, 385, 387, 388, 389, 390, 392, 393, 394, 395, 396, 397, 398, 399, 411, 412, 413, 414, 415, 416, 417, 419, 420, 421, 435, 440, 442, 443, 444, 445, 446, 447, 449, 454, 463, 476, 478, 480, 482, 483, 484, 486], "functool": 479, "further": [2, 8, 481], "fuse": [1, 479], "fusibl": 479, "futur": [5, 381, 482, 483], "g": [3, 8, 112, 143, 189, 237, 349, 448, 466, 467, 483, 488], "g_t": [349, 455, 457, 458, 459, 460, 461, 466, 467], "gain": [402, 403, 404, 405], "gamma": [328, 344, 348, 350, 382, 402, 403, 404, 405], "gap": 1, "gate": [342, 343, 414], "gather": [0, 121, 163, 164], "gather_mm": [0, 164], "gather_qmm": 0, "gaurante": 301, "gaussian": [4, 341, 411, 412, 413, 424], "gaussian_nll_loss": 324, "gelu": [324, 412, 413, 479], "gelu_approx": [324, 341, 411], "gelu_fast_approx": [324, 341, 411], "geluapprox": 341, "gelufast": 341, "gener": [0, 1, 2, 3, 4, 11, 18, 101, 140, 143, 154, 155, 193, 208, 240, 245, 246, 247, 248, 251, 252, 399, 476, 479, 482, 483, 488], "general_": 2, "generate_stub": 8, "geq": [397, 446], "get": [2, 4, 6, 8, 99, 100, 101, 103, 104, 113, 114, 210, 211, 212, 213, 243, 324, 479, 481, 483, 487], "get_cache_memori": 209, "get_command_encod": 2, "get_kernel": 2, "gguf": [8, 194, 263, 486], "gh": 1, "gii": 1, "git": 8, "github": [4, 6, 8, 479], "give": [2, 5, 6, 28, 479], "given": [0, 2, 8, 15, 17, 28, 38, 83, 91, 93, 95, 108, 109, 110, 111, 116, 118, 131, 136, 138, 141, 147, 148, 149, 150, 151, 152, 156, 157, 158, 162, 163, 189, 203, 205, 207, 216, 221, 225, 227, 235, 245, 247, 248, 259, 260, 268, 273, 275, 280, 284, 286, 292, 293, 294, 296, 297, 298, 300, 315, 326, 327, 336, 355, 356, 362, 378, 422, 424, 430], "gix": 1, "gix_mult": 1, "giy_mult": 1, "global": [119, 121, 122, 123, 125, 126, 127, 132, 249, 309, 476, 479], "glorot": [402, 403], "glorot_norm": 324, "glorot_uniform": 324, "glu": [5, 324], "gm": 1, "gn": 1, "go": [2, 5, 481], "golub": 189, "good": [2, 8, 454, 479, 480, 487], "goroshin": 337, "gower": 5, "gpu": [1, 3, 7, 8, 210, 482, 487], "gputrac": [3, 219], "grad": [2, 4, 6, 299, 309, 454, 462, 479, 480, 481, 482, 483, 485], "grad_fn": [4, 479, 481], "gradient": [0, 4, 6, 112, 165, 281, 299, 308, 309, 324, 363, 376, 381, 399, 429, 452, 454, 455, 456, 458, 459, 460, 461, 462, 465, 467, 479, 480, 481, 482, 483, 484, 485], "grain": 476, "graph": [2, 5, 6, 7, 481], "great": 3, "greater": [0, 5, 28, 139, 167, 233, 309, 397, 446], "greater_equ": 0, "grep": 8, "grid": [2, 143, 208], "grid_dim": 2, "grid_grad": 1, "grid_idx": 1, "grid_sampl": 1, "grid_sample_grad": 1, "grid_sample_ref": 1, "grid_sample_vjp": 1, "grid_shap": 1, "grid_siz": 1, "ground": [4, 5, 423, 433], "group": [0, 1, 98, 99, 100, 101, 102, 103, 104, 116, 121, 122, 123, 125, 126, 127, 141, 146, 164, 237, 238, 301, 307, 330, 344, 380, 381, 480], "group_dim": 2, "group_siz": [0, 116, 141, 164, 237, 238, 307, 380, 381], "groupnorm": 324, "grow": 483, "gru": 324, "guid": [2, 7], "gw": 1, "h": [1, 2, 98, 99, 100, 102, 103, 104, 189, 327, 328, 331, 332, 334, 335, 337, 338, 343, 349, 356, 383, 481, 483], "h_": [327, 343, 349, 356, 383], "h_in": 1, "h_stride": 1, "h_t": [343, 349, 383], "ha": [2, 3, 5, 6, 7, 8, 78, 94, 118, 127, 153, 154, 156, 157, 158, 165, 183, 184, 186, 187, 188, 191, 192, 208, 212, 241, 328, 343, 349, 352, 383, 452, 454, 479, 482, 483, 485, 487], "had": 5, "hadamard": [0, 168], "hadamard_transform": 0, "half": [2, 18, 248, 252, 386, 483], "halv": [342, 414], "hand": [5, 481, 483], "handi": 481, "handl": [2, 324, 479], "happen": [2, 5, 142, 399, 454, 479, 483], "happi": 5, "hard": 5, "hard_shrink": [324, 345], "hard_tanh": [324, 346], "hardshrink": [324, 415], "hardswish": 324, "hardtanh": [324, 416], "hat": [116, 141, 237], "have": [0, 1, 2, 5, 8, 16, 82, 84, 85, 86, 90, 121, 154, 155, 157, 158, 164, 172, 204, 219, 241, 301, 310, 349, 378, 388, 461, 463, 478, 479, 480, 482, 483, 487], "haven": 5, "hazan": 457, "he": [5, 404, 405], "he_norm": 324, "he_uniform": 324, "head": [146, 378, 399], "header": [2, 143], "heart": 5, "heavi": 5, "height": [327, 328, 331, 332, 334, 335, 337, 338, 356], "hello": [310, 314], "help": [2, 5, 479, 487], "helper": [5, 143, 479], "henc": [0, 2, 237, 479], "hendryck": 413, "here": [2, 5, 454, 479, 481, 483, 486, 487], "hermitian": [186, 187], "hf": 349, "hg": 349, "hh": 383, "hi": [5, 349], "hidden": [343, 349, 383, 399], "hidden_dim": [6, 452, 454], "hidden_s": [343, 349, 383], "hierarchi": 317, "high": [248, 252, 324, 340, 408, 449], "high_pad_s": 0, "higher": [2, 171, 218, 430, 481], "highli": 8, "him": 5, "hing": 425, "hinge_loss": 324, "hinton": 466, "hit": 2, "hn": 343, "ho": 349, "hold": [2, 5, 10, 11, 189, 479], "homebrew": 480, "hopkin": 189, "host": 2, "host1": 480, "host2": 480, "host_nam": [1, 2], "hostfil": 480, "hostnam": 480, "hot": 423, "hour": 5, "how": [2, 5, 6, 324, 326, 327, 330, 331, 332, 333, 334, 335, 340, 355, 356, 380, 400, 462, 479, 482, 487], "howev": [2, 112, 324, 341, 344, 463, 476, 479, 480, 483, 484], "hr": 343, "http": [344, 348, 350, 357, 382, 413, 435], "huber": 426, "huber_loss": 324, "human": [404, 405], "hundr": 8, "hurri": 5, "hutter": 459, "hyperbol": [0, 20, 22, 25, 107, 272, 290, 398, 447], "hz": 343, "i": [0, 1, 2, 3, 5, 6, 7, 8, 16, 18, 28, 37, 78, 83, 93, 99, 100, 101, 103, 104, 105, 108, 109, 110, 111, 112, 117, 118, 121, 122, 124, 125, 126, 127, 129, 136, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 161, 162, 163, 164, 168, 172, 173, 178, 179, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 194, 199, 203, 204, 208, 214, 217, 218, 232, 233, 236, 237, 238, 245, 246, 247, 257, 259, 262, 263, 264, 269, 273, 275, 280, 281, 286, 287, 288, 291, 294, 295, 299, 300, 301, 302, 303, 304, 307, 309, 310, 311, 312, 313, 317, 319, 324, 326, 327, 328, 330, 331, 332, 333, 334, 335, 336, 337, 338, 341, 343, 344, 348, 349, 350, 352, 355, 356, 362, 363, 369, 371, 372, 374, 375, 377, 378, 379, 381, 382, 383, 386, 391, 397, 399, 400, 404, 405, 411, 413, 421, 422, 424, 429, 430, 433, 434, 436, 441, 446, 452, 454, 456, 459, 461, 462, 463, 468, 470, 471, 476, 479, 480, 481, 482, 483, 484, 485, 486, 487, 488], "i386": 8, "i_n": 1, "i_nw": 1, "i_s": 1, "i_sw": 1, "i_t": 349, "iclr": [458, 459, 460], "id": [6, 8], "idea": [481, 483], "idempot": [363, 375], "ident": [0, 112, 140, 281, 324, 372], "identifi": [2, 310, 478], "idim": 6, "idiom": [6, 479], "idx": [38, 482], "ie": [375, 480], "ieee": 317, "ignor": [5, 38, 93, 94, 136, 456], "ih": 383, "ii": 1, "ij": 208, "imag": [0, 331, 332, 334, 335, 337, 338, 400], "imagenet": [404, 405], "imaginari": 170, "immedi": [5, 358], "implement": [0, 1, 4, 6, 145, 146, 189, 340, 362, 378, 386, 388, 391, 397, 399, 400, 446, 455, 456, 457, 458, 460, 461, 462, 474, 479, 481], "impli": 301, "implicit": [476, 479, 481], "implicitli": 483, "import": [2, 3, 4, 5, 6, 8, 112, 168, 189, 265, 299, 310, 311, 312, 313, 314, 324, 326, 327, 328, 348, 355, 356, 365, 400, 421, 423, 430, 449, 452, 454, 479, 480, 481, 482, 483, 484, 485], "improv": [1, 2, 3, 5, 421, 455, 456, 457, 458, 459, 460, 466, 479, 480], "in_ax": [303, 481], "in_channel": [330, 331, 332, 333, 334, 335], "in_dim": [324, 452], "in_proj": 452, "inci": 2, "includ": [1, 2, 108, 109, 110, 111, 143, 211, 212, 217, 350, 359, 371, 381, 424, 454, 479, 481, 482, 485, 486, 488], "include_dir": 2, "inclus": [0, 41, 42, 43, 44, 108, 109, 110, 111, 159], "incom": 2, "inconveni": 479, "incorpor": 484, "incorrect": 484, "increas": 218, "increment": 18, "incur": [5, 8], "incx": 2, "independ": [120, 337, 338], "index": [0, 1, 2, 7, 9, 28, 38, 138, 140, 165, 208, 233, 287, 288, 299, 315], "indic": [0, 2, 16, 26, 27, 28, 29, 38, 163, 164, 165, 172, 173, 174, 175, 176, 177, 178, 191, 236, 275, 287, 288, 299, 372, 374, 423, 430, 470, 482], "indices_or_sect": [71, 275], "indirectli": 484, "individu": [324, 337, 338], "ineffici": [482, 483], "inexact": [11, 178], "inf": [189, 225, 378], "infer": [7, 162, 194, 294, 480], "infin": [0, 174, 176, 177, 225, 355, 356, 460], "infinit": [16, 172, 173], "info": [5, 8], "inform": [3, 5, 6, 8, 131, 210, 263, 264, 317, 324, 328, 341, 378, 481, 487], "inherit": [6, 478], "inifn": 174, "init": [324, 379, 449, 454, 468, 469, 471, 472, 480], "init_fn": [401, 402, 403, 404, 405, 406, 407, 408, 449], "init_valu": 1, "initi": [1, 3, 4, 5, 123, 313, 324, 328, 344, 348, 350, 352, 379, 382, 401, 402, 403, 404, 405, 406, 407, 408, 452, 463, 468, 469, 471, 472, 479, 480, 483], "initializer_list": 0, "inject": 0, "inlin": 0, "inner": [0, 479], "inorm": 348, "inp": [1, 143], "inp_ndim": 1, "inp_shap": 1, "inp_strid": 1, "inplac": [2, 8], "input": [0, 1, 2, 4, 5, 12, 13, 14, 15, 16, 17, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 115, 117, 118, 121, 122, 127, 128, 129, 130, 131, 133, 134, 135, 137, 138, 139, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 163, 164, 165, 166, 167, 168, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 221, 222, 223, 224, 225, 226, 228, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 247, 250, 253, 254, 255, 256, 257, 258, 259, 260, 261, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 283, 284, 285, 287, 288, 289, 290, 291, 292, 293, 294, 295, 297, 298, 299, 300, 301, 303, 304, 306, 326, 327, 328, 330, 331, 332, 333, 334, 335, 337, 338, 340, 342, 343, 344, 348, 349, 350, 352, 355, 356, 378, 381, 382, 383, 386, 397, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 414, 421, 422, 424, 425, 426, 427, 429, 430, 432, 434, 446, 449, 479, 481, 482, 485, 486], "input_dil": [0, 101], "input_dim": [6, 324, 352, 381], "input_nam": [1, 143], "input_s": [343, 349, 383], "inputs1": 430, "inputs2": 430, "insert": [118, 138, 487], "insid": 479, "inspect": [3, 479, 485], "inspir": 7, "instabl": 434, "instal": 2, "instanc": [5, 38, 112, 237, 314, 324, 348, 358, 359, 360, 363, 365, 366, 367, 372, 375, 376, 377, 388, 452, 484], "instancenorm": 324, "instanti": [1, 2, 6, 483], "instantiate_axpbi": 2, "instead": [2, 8, 112, 324, 377, 391, 480, 481, 483], "int": [0, 1, 2, 5, 6, 9, 15, 17, 18, 26, 27, 28, 29, 33, 34, 35, 36, 41, 42, 43, 44, 45, 46, 49, 56, 57, 58, 59, 60, 63, 66, 68, 71, 74, 75, 76, 77, 78, 80, 83, 90, 91, 95, 98, 99, 100, 101, 102, 103, 104, 108, 109, 110, 111, 116, 117, 118, 125, 126, 127, 131, 138, 140, 141, 145, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 162, 164, 165, 169, 178, 185, 189, 193, 203, 205, 207, 210, 211, 212, 213, 216, 217, 218, 221, 223, 229, 232, 233, 235, 236, 237, 238, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 256, 257, 259, 260, 273, 274, 275, 278, 279, 280, 284, 285, 287, 288, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 303, 305, 307, 315, 324, 326, 327, 328, 330, 331, 332, 333, 334, 335, 340, 342, 343, 344, 348, 349, 350, 352, 355, 356, 378, 380, 381, 382, 383, 386, 391, 399, 414, 422, 423, 427, 432, 434, 452, 468, 470, 471, 472], "int16": 317, "int32": [0, 1, 11, 18, 38, 159, 178, 189, 248, 317, 400, 482, 485], "int64": [11, 317], "int8": [11, 317], "int_0": 134, "integ": [0, 11, 161, 163, 164, 178, 189, 210, 232, 237, 238, 240, 247, 248, 275, 287, 291, 303, 317, 340, 370, 470, 482], "integr": [18, 287, 483], "intend": [0, 479], "interact": 399, "interest": 487, "interfac": 2, "intermedi": 484, "intern": 328, "interpol": 400, "interv": [18, 193, 248, 252], "introduc": [0, 259], "intuit": 324, "invalid": [0, 83], "invers": [0, 19, 20, 21, 22, 23, 24, 25, 135, 150, 151, 152, 153, 154, 155, 184, 188, 192], "invert": 0, "involv": [454, 479], "iogpu": 218, "ip": 480, "is_avail": 123, "is_equival": 2, "is_floating_point": 2, "is_leaf": [310, 311, 312, 313], "is_leaf_fn": 362, "isclos": 0, "isfinit": 0, "ish": 5, "ishmael": 5, "isinf": 0, "isnan": 0, "isneginf": 0, "isposinf": 0, "issu": [480, 481, 484], "issubdtyp": [11, 317], "item": [0, 2, 4, 5, 6, 311, 454, 483, 484, 485], "iter": [4, 6, 191, 311, 312, 476, 479, 483], "iterm": 8, "itertool": [5, 311], "its": [0, 1, 2, 8, 184, 204, 233, 250, 296, 308, 314, 324, 381, 454, 458, 459, 460, 480, 483, 484, 487], "itself": [2, 307, 463], "ix": 1, "ix_n": 1, "ix_nw": 1, "ix_s": 1, "ix_sw": 1, "iy_n": 1, "iy_nw": 1, "iy_s": 1, "iy_sw": 1, "j": [5, 8, 189, 337, 457, 458, 460], "j8": 2, "jacobian": [2, 179, 302, 485], "jain": 337, "jax": [7, 476], "jit": 143, "jmlr": 457, "jnp": 484, "john": 189, "join": 470, "join_schedul": 454, "jointli": 245, "just": [2, 6, 350, 479, 482], "jvp": [2, 112, 485], "k": [0, 5, 45, 90, 117, 140, 146, 163, 168, 293, 296, 297, 298, 326, 352, 355, 363], "k_h": [327, 356], "k_w": [327, 356], "kaim": 405, "keep": [2, 15, 17, 26, 27, 203, 205, 207, 221, 235, 280, 284, 300, 324, 362, 481, 483], "keepdim": [0, 15, 17, 26, 27, 33, 34, 35, 36, 56, 57, 58, 59, 63, 75, 76, 80, 189, 203, 205, 207, 221, 235, 273, 280, 284, 300], "kei": [1, 3, 5, 146, 210, 240, 241, 242, 244, 245, 246, 247, 248, 250, 251, 252, 310, 311, 362, 363, 375, 378, 463, 476, 478, 481], "kept": 218, "kernel": [2, 7, 8, 98, 99, 100, 101, 102, 103, 104, 143, 326, 327, 355, 356, 479, 482], "kernel_dil": [0, 101], "kernel_s": [326, 327, 330, 331, 332, 333, 334, 335, 355, 356], "key_cach": 5, "key_input_dim": 378, "key_proj": 5, "keyword": [165, 265, 266, 299, 311, 324, 476, 486, 488], "kind": 5, "kingma": [458, 460], "kl_div_loss": 324, "kname": 2, "know": [2, 5], "known": [389, 441], "kth": [0, 28, 233], "kullback": 427, "kw_onli": 2, "kwarg": [10, 120, 265, 266, 488], "l": [5, 6, 183, 184, 186, 187, 324, 326, 328, 330, 333, 343, 349, 355, 383, 433], "l1": [299, 426, 428, 429, 433], "l1_loss": 324, "l2": [426, 429, 467], "l2_loss": 324, "l_": [326, 355, 426], "la": 189, "label": [3, 4, 423, 430], "label_smooth": 423, "lack": 482, "lambd": [345, 395, 415, 445], "lambda": [311, 312, 313, 324, 345, 358, 363, 370, 395, 415, 439, 445, 455, 456, 457, 458, 459, 460, 461, 466, 467, 479, 480, 481], "languag": [1, 2], "larg": [5, 324, 378, 429, 479, 480, 483], "larger": [1, 145, 218, 386, 461], "largest": [189, 225, 293], "lasso": 299, "last": [0, 1, 5, 29, 78, 142, 144, 149, 152, 154, 155, 157, 158, 159, 163, 164, 171, 183, 184, 186, 187, 188, 190, 191, 192, 204, 213, 241, 274, 291, 301, 330, 331, 332, 333, 334, 335, 337, 338, 344, 400, 484], "latenc": 480, "later": [3, 8, 454], "launch": [1, 2, 123, 480, 482], "layer": [7, 142, 307, 324, 326, 327, 337, 338, 343, 344, 349, 350, 352, 355, 356, 372, 377, 380, 381, 383, 388, 399, 448, 452], "layer_s": 6, "layernorm": 324, "layout": 1, "lazi": [7, 452, 485], "lazili": [5, 324], "lceil": 90, "ld": [343, 349, 383], "ldot": [326, 327, 355, 356], "lead": [0, 18, 83, 479], "leaf": [94, 307, 310, 311, 312, 313, 362], "leaf_modul": 324, "leaki": [351, 418], "leaky_relu": 324, "leakyrelu": 324, "learn": [4, 6, 7, 328, 344, 348, 350, 379, 382, 454, 455, 456, 457, 458, 459, 460, 461, 466, 467], "learnabl": [330, 331, 332, 333, 334, 335, 388], "learning_r": [6, 454, 455, 456, 457, 458, 459, 460, 461, 463, 466, 467, 468, 469, 470, 471, 472, 479], "least": [5, 84, 85, 86, 93, 183, 184, 186, 187, 188, 190, 191, 192, 237], "leav": [2, 136, 311, 312, 313], "lectur": 466, "lecun": 337, "left": [0, 5, 145, 180, 189, 237, 259, 326, 327, 341, 355, 356, 386, 400, 412, 413, 424, 426, 434], "left_shift": 0, "leibler": 427, "len": [5, 149, 152, 155, 158, 168, 470], "length": [5, 278, 328, 330, 333, 343, 349, 383, 470], "leq": [426, 439], "less": [0, 1, 5, 28, 182, 218, 233, 386, 433], "less_equ": 0, "let": [1, 2, 4, 5, 184, 479, 481, 483, 484], "level": [0, 163, 164, 404, 405], "lfloor": [326, 327, 355, 356], "lh": [343, 349, 383], "lhs_indic": [0, 163, 164], "lhs_mask": 90, "lib": 480, "libmlx": 8, "libmlx_ext": 2, "libmpi": 480, "librari": [2, 8, 319, 324], "like": [2, 5, 7, 126, 178, 230, 306, 338, 429, 463, 465, 479, 480, 481, 483, 484, 485, 487], "likelihood": [424, 432], "limit": [0, 2, 93, 216, 217, 218, 482], "linalg": 168, "line": [5, 480, 483, 484], "linear": [0, 2, 5, 6, 7, 307, 311, 324, 329, 339, 341, 342, 351, 365, 381, 383, 384, 385, 387, 389, 400, 409, 410, 411, 412, 413, 414, 418, 437, 438, 439, 441, 449, 452, 463, 471, 479], "linear1": 5, "linear2": 5, "linear3": 5, "linear_schedul": [454, 470], "linearli": 378, "link": [2, 8], "linspac": 0, "lion": 454, "list": [1, 5, 10, 15, 17, 30, 71, 78, 83, 84, 85, 86, 91, 94, 95, 101, 131, 136, 143, 148, 149, 151, 152, 154, 155, 157, 158, 162, 165, 179, 189, 203, 205, 207, 208, 221, 229, 232, 235, 240, 241, 242, 244, 245, 246, 248, 251, 252, 263, 273, 275, 279, 280, 284, 291, 292, 295, 299, 300, 302, 305, 310, 313, 314, 324, 363, 365, 366, 367, 368, 373, 375, 376, 377, 452, 454, 458, 459, 460, 461, 470, 478, 479, 480, 481, 483], "liter": [2, 232, 400, 404, 405, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434], "littl": 5, "liu": 5, "live": [7, 143, 487], "ll": [1, 4, 6, 426, 479, 481], "llama": 5, "llamaattent": 5, "llamaencoderlay": 5, "llm": 7, "load": [6, 7, 319, 365, 480], "load_weight": [324, 483], "loader": 6, "loader_path": 2, "loan": 189, "loc": [1, 244, 246], "local": [324, 337, 480], "locat": [0, 2, 83, 376, 377, 480, 487], "log": [0, 197, 199, 203, 353, 354, 419, 420, 421, 424, 427, 429, 432, 444], "log10": 0, "log1p": 0, "log2": 0, "log_cosh_loss": 324, "log_sigmoid": [324, 353], "log_softmax": [324, 354], "logaddexp": 0, "logarithm": [0, 195, 196, 197, 198], "logcosh": 429, "logic": [0, 2, 200, 201, 202], "logical_and": 0, "logical_not": 0, "logical_or": 0, "logist": [0, 4, 269, 413, 441], "logit": [5, 241, 421, 423, 479], "logsigmoid": 324, "logsoftmax": 324, "logsumexp": 0, "long": 5, "longer": [5, 105, 481], "look": [2, 5, 480], "lookup": 340, "loop": [5, 6, 479, 480, 481, 483], "loshchilov": 459, "loss": [4, 6, 299, 324, 454, 479, 480, 481, 483], "loss_and_grad": 324, "loss_and_grad_fn": [6, 454, 479, 481], "loss_fn": [4, 6, 454, 479, 481], "loss_grad_fn": 480, "lot": [480, 481], "low": [248, 252, 408, 449], "low_pad_s": 0, "lower": [183, 184, 186, 187, 192, 237, 248, 251, 252, 296, 408], "lr": [4, 461], "lr_schedul": [468, 469, 470, 472], "lstm": 324, "lto": 2, "lu": 5, "luckili": 483, "lvalu": 299, "m": [0, 2, 5, 8, 90, 140, 163, 168, 189, 296, 326, 327, 355, 356, 455, 479], "m1": [1, 5, 479, 481, 487], "m10": 317, "m7": 317, "m_": [458, 459, 460, 461], "m_t": [458, 459, 460, 461], "mac": 480, "machin": [5, 7, 8, 466, 480], "maco": [8, 218], "macosx": 8, "made": [5, 319], "mai": [2, 189, 307, 337, 480, 481, 482], "main": [7, 118, 140, 143, 294, 311, 312, 324, 480], "maintain": [337, 338, 461], "major": [0, 2], "make": [1, 2, 3, 5, 6, 8, 204, 227, 268, 324, 468, 469, 471, 472, 479, 483, 485, 487], "make_shar": 2, "malloc_or_wait": 2, "man": 5, "manag": [282, 476, 480, 487], "mani": [2, 83, 275, 330, 331, 332, 333, 334, 335, 340, 380, 479, 480, 483], "manual": 324, "map": [2, 6, 38, 194, 311, 340, 358], "map_fn": [358, 362], "map_torch_to_mlx": 5, "margin": [430, 434], "margin_ranking_loss": 324, "mask": [0, 5, 90, 146, 372, 378, 482], "mask_lh": [0, 90], "mask_n": 1, "mask_nw": 1, "mask_out": [0, 90], "mask_rh": [0, 90], "mask_s": 1, "mask_sw": 1, "matadata": 194, "match": [8, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 211, 365, 400, 423, 480, 482, 484], "materi": [5, 7], "math": [5, 434, 479], "mathbf": 184, "mathcal": 352, "mathemat": 189, "mathrm": [134, 269, 348], "matmul": [0, 163, 487], "matric": [189, 190, 191], "matrix": [0, 4, 14, 45, 90, 116, 117, 140, 141, 163, 164, 168, 169, 183, 184, 186, 187, 188, 189, 190, 191, 192, 204, 208, 237, 238, 245, 380, 381, 406, 449], "matter": [5, 324], "max": [0, 1, 2, 189, 206, 329, 355, 356, 379, 409, 416, 417, 422, 424, 425, 430, 434, 436, 438, 456, 460, 479, 481, 487], "max_": [355, 356], "max_buffer_s": 210, "max_freq": 391, "max_i": 237, "max_norm": 309, "max_recommended_working_set_s": [210, 218], "max_val": 416, "maximum": [0, 6, 26, 38, 93, 108, 213, 217, 309, 324, 351, 384, 391, 412, 413, 418, 437, 452, 483], "maxpool1d": 324, "maxpool2d": 324, "maxtotalthreadsperthreadgroup": 2, "mca": 480, "md": 189, "me": 5, "mean": [0, 1, 4, 5, 6, 144, 244, 245, 246, 299, 324, 328, 344, 363, 382, 407, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 479, 481, 484], "meant": 112, "measur": 487, "mechan": 399, "medic": 338, "meet": 8, "member": [2, 324, 368, 373], "memori": [0, 1, 2, 7, 83, 209, 211, 212, 213, 215, 216, 217, 218, 399, 452, 456, 479, 483, 484], "memory_order_relax": 1, "memory_s": [210, 218], "memoryview": [483, 484], "merg": 479, "meshgrid": 0, "metadata": [4, 194, 263, 264], "metal": [2, 7, 143], "metal_captur": 3, "metal_kernel": 1, "metal_path": 8, "metallib": [2, 8], "method": [2, 5, 9, 10, 30, 112, 120, 307, 315, 324, 325, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350, 351, 352, 353, 354, 355, 356, 357, 369, 378, 379, 380, 381, 382, 383, 384, 385, 386, 387, 388, 389, 390, 391, 392, 393, 394, 395, 396, 397, 398, 399, 400, 452, 455, 456, 457, 458, 459, 460, 461, 463, 466, 467, 474], "millisecond": [8, 479, 487], "min": [0, 2, 189, 222, 329, 379, 409, 416, 417, 436, 438], "min_freq": 391, "min_i": 237, "min_val": 416, "mind": [2, 5], "mine": 5, "minibatch": 6, "minim": 480, "minimum": [0, 27, 38, 93, 109, 391, 421, 422], "minsizerel": 8, "minu": 139, "minut": 5, "mish": 324, "miss": [365, 486], "mix": 482, "mkdir": [3, 8], "ml": 8, "mlp": [6, 324, 399, 454], "mlp_dim": [5, 399], "mlx": [1, 3, 4, 5, 6, 8, 319, 324, 449, 452, 454, 476, 478, 479, 480, 481, 482, 483, 484, 485, 486, 487], "mlx_build_benchmark": 8, "mlx_build_cpu": 8, "mlx_build_exampl": 8, "mlx_build_gguf": 8, "mlx_build_met": [2, 8], "mlx_build_metallib": 2, "mlx_build_python_bind": 8, "mlx_build_safetensor": 8, "mlx_build_test": 8, "mlx_disable_compil": [119, 132, 479], "mlx_ext": 2, "mlx_ext_metallib": 2, "mlx_include_dir": 2, "mlx_metal_debug": [3, 8], "mlx_metal_jit": 8, "mlx_sample_extens": 2, "mlx_trace": 3, "mnist": 6, "mode": [0, 1, 2, 105, 232, 361, 372, 374, 400, 404, 405, 480], "model": [4, 6, 7, 265, 307, 308, 311, 312, 324, 358, 361, 363, 365, 369, 372, 374, 375, 376, 378, 399, 449, 452, 454, 462, 463, 465, 479, 480, 483], "modest": 2, "modif": 484, "modifi": 484, "modul": [2, 5, 6, 307, 308, 388, 399, 449, 465, 478, 479, 483], "moment": [5, 456, 458, 459, 460], "momentum": [328, 461, 463, 467, 479], "monei": 5, "monoton": 435, "more": [1, 2, 3, 6, 10, 78, 118, 163, 183, 184, 186, 187, 188, 191, 192, 204, 216, 217, 263, 264, 317, 324, 328, 337, 386, 391, 399, 400, 402, 403, 404, 405, 421, 476, 479, 480, 481, 482, 485, 487], "most": [2, 241, 324, 465, 479, 480, 481, 482, 483], "move": [0, 2, 223, 487], "moveaxi": 0, "mpi": 319, "mpiexec": 480, "mpirun": 480, "mse": 299, "mse_loss": 324, "mtl": 2, "mtl_capture_en": 3, "mtlcommandbuff": 2, "mu": 467, "much": [1, 2, 5, 326, 327, 355, 356, 479, 483], "multi": [7, 146, 330, 331, 332, 333, 334, 335, 482, 484], "multidimension": 208, "multiheadattent": [5, 324], "multipl": [0, 1, 8, 14, 90, 142, 144, 163, 164, 204, 224, 237, 238, 378, 391, 469, 470, 472, 479, 483, 486], "multipli": [0, 2, 38, 164, 237, 238, 336, 391, 400], "murtadha": 5, "must": [0, 1, 2, 3, 8, 90, 93, 145, 162, 164, 186, 187, 189, 240, 241, 245, 248, 251, 252, 304, 400, 484], "mx": [1, 2, 3, 4, 5, 6, 38, 96, 97, 112, 123, 126, 143, 159, 178, 186, 187, 189, 190, 194, 247, 265, 299, 309, 324, 326, 327, 328, 339, 348, 351, 355, 356, 358, 365, 369, 384, 400, 401, 402, 403, 404, 405, 406, 407, 408, 410, 418, 421, 422, 423, 427, 430, 437, 447, 449, 452, 454, 476, 479, 480, 481, 482, 483, 484, 485, 486, 487, 488], "my": [5, 8], "my_devic": 488, "my_path": 265, "myexp": [1, 143], "myexp_strid": 1, "mymlp": 452, "n": [0, 1, 2, 5, 30, 90, 98, 99, 100, 101, 102, 103, 104, 140, 147, 149, 150, 152, 153, 156, 158, 168, 169, 245, 280, 296, 300, 326, 327, 328, 330, 331, 332, 333, 334, 335, 337, 338, 343, 349, 355, 356, 383, 400, 429, 434, 480], "n_i": [326, 327, 355, 356], "n_t": 343, "naiv": [2, 481], "naive_add": 481, "name": [1, 2, 143, 164, 194, 237, 238, 263, 264, 265, 266, 324, 344, 362, 365, 367, 480, 482, 486], "named_modul": 324, "nan": [0, 16, 82, 172, 173, 175, 225], "nan_to_num": 0, "nanobind": [2, 399], "nanobind_add_modul": 2, "nativ": 8, "natur": [0, 195, 197, 483], "nb": 2, "nb_domain": 2, "nb_func": 399, "nb_modul": 2, "nb_static": 2, "nbyte": 2, "nc": 328, "ndarrai": [30, 482, 483, 485], "ndhwc": [332, 335, 338], "ndim": [0, 1, 2, 159, 189, 191, 400], "ne": 1, "nearest": [1, 400], "necessari": 324, "necessarili": 293, "need": [1, 2, 5, 6, 7, 8, 82, 237, 324, 376, 377, 391, 399, 476, 480, 481, 483, 484, 485, 487], "neg": [0, 118, 159, 176, 225, 259, 294, 351, 355, 356, 378, 424, 432, 434, 482], "negat": [0, 226], "negative_slop": [351, 418], "neginf": [0, 225], "neighbor": 400, "neither": [165, 299], "nelem": 2, "nervou": 5, "nest": [78, 94, 313, 324, 452, 478, 481], "nesterov": 467, "network": [5, 7, 328, 337, 340, 402, 403, 449, 452, 466, 480], "neural": [5, 7, 340, 402, 403, 435, 449, 452, 466], "never": [5, 483], "new": [0, 2, 6, 91, 118, 223, 227, 257, 279, 295, 301, 311, 312, 370, 378, 452, 454, 465, 470, 479, 482, 483, 484], "new_tre": 312, "next": [2, 5, 6, 216], "nh": [343, 349, 383], "nhwc": [328, 331, 334], "nice": [481, 483], "nlc": [328, 330, 333], "nld": [343, 349, 383], "nlh": [343, 349, 383], "nll": [424, 432], "nll_loss": 324, "nn": [2, 5, 6, 265, 311, 324, 449, 452, 454, 463, 465, 479, 483], "nobodi": 5, "node": [94, 136, 303, 312, 313], "nois": 4, "noisi": 4, "nomins": 2, "non": [0, 1, 2, 8, 208, 373, 383, 435, 452], "none": [1, 2, 5, 9, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 115, 116, 117, 118, 119, 121, 122, 125, 126, 127, 128, 129, 130, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 215, 219, 220, 221, 222, 223, 224, 225, 226, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 264, 265, 267, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 303, 304, 305, 306, 307, 310, 311, 312, 313, 315, 326, 327, 341, 355, 356, 358, 362, 363, 370, 375, 378, 383, 391, 399, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 456, 474, 482], "nonlinear": [383, 479], "nonzero": 482, "noop": 375, "nor": [2, 165, 299], "norm": [5, 144, 309, 344, 434, 460, 461], "norm1": 5, "norm2": 5, "norm_first": 399, "normal": [1, 2, 4, 5, 142, 143, 144, 186, 245, 251, 324, 326, 327, 328, 344, 348, 350, 355, 356, 382, 399, 402, 404, 484, 487], "not_equ": 0, "notabl": [5, 7], "notat": [116, 141, 310, 367], "note": [0, 1, 2, 5, 8, 16, 18, 83, 90, 94, 100, 103, 104, 112, 146, 154, 155, 164, 172, 189, 211, 237, 241, 301, 307, 324, 382, 400, 454, 484, 486], "noth": [5, 324, 483], "notic": [5, 481, 486], "now": [1, 2, 5, 8, 381, 479, 480, 484], "np": [1, 5, 6, 480, 484, 485], "npy": [194, 262, 486], "npz": [5, 194, 265, 266, 365, 369, 486], "nuclear": 189, "nullopt": 0, "num": [0, 5, 193, 250], "num_class": [6, 454], "num_decoder_lay": 399, "num_embed": [340, 380], "num_encoder_lay": 399, "num_epoch": [6, 454], "num_exampl": 4, "num_featur": [4, 328], "num_group": 344, "num_head": [5, 378, 399], "num_it": 4, "num_lay": [5, 6, 454], "num_param": 324, "num_paramet": 379, "num_sampl": 241, "num_split": 0, "number": [0, 2, 11, 18, 61, 70, 94, 99, 100, 101, 103, 104, 116, 140, 141, 164, 165, 169, 179, 193, 225, 232, 237, 238, 241, 244, 246, 250, 252, 256, 259, 260, 291, 292, 296, 299, 302, 303, 307, 324, 328, 330, 331, 332, 333, 334, 335, 337, 338, 344, 348, 378, 379, 399, 400, 402, 403, 404, 405, 468, 470, 471, 476, 479, 481, 488], "number_of_el": 0, "numer": [5, 142, 144, 189, 199, 203, 273, 328, 344, 348, 350, 382, 421, 422, 424, 434, 455, 456, 457, 458, 459, 460, 466, 479, 483], "numpi": [2, 5, 6, 7, 13, 16, 18, 87, 88, 89, 91, 128, 129, 133, 166, 167, 172, 180, 181, 182, 199, 204, 206, 222, 224, 228, 234, 255, 258, 283, 483, 485, 486], "nw": 1, "nwhc": 337, "o": [2, 8, 146, 349], "o_t": 349, "obj": 263, "object": [3, 10, 30, 50, 78, 94, 143, 178, 265, 303, 310, 311, 312, 313, 317, 337, 399, 478], "observ": 5, "occupi": [116, 141, 164, 237, 238], "occur": 484, "odim": 6, "odot": [343, 349], "off": [5, 8, 483], "offer": 429, "offset": [0, 1, 2, 5, 46, 83, 118, 142, 145, 294], "often": 338, "ok": [365, 481], "okai": [479, 483], "old": 5, "omit": [458, 460, 480], "onc": [2, 8, 479], "one": [0, 2, 5, 8, 38, 78, 84, 93, 99, 100, 101, 103, 104, 138, 140, 142, 144, 145, 189, 197, 204, 238, 241, 278, 283, 317, 375, 400, 423, 480, 487], "ones": [0, 2, 5, 230, 265, 296, 376, 377, 454, 480, 482], "ones_lik": 0, "onli": [1, 2, 5, 7, 8, 82, 90, 99, 100, 101, 103, 104, 186, 187, 189, 218, 237, 245, 301, 324, 362, 363, 365, 370, 372, 375, 376, 377, 452, 479, 480, 481, 486, 487], "onlin": 457, "op": [1, 2, 231, 301, 363, 483], "open": [3, 8, 18, 248, 252], "openmpi": 480, "oper": [3, 5, 7, 9, 37, 84, 85, 86, 101, 146, 163, 164, 234, 236, 273, 281, 288, 315, 324, 399, 461, 479, 480, 481, 482, 483, 484, 485, 487, 488], "operand": [130, 131, 163], "opportun": 479, "opt": [462, 480], "optim": [1, 3, 4, 6, 7, 376, 479, 480, 481, 483], "option": [0, 3, 5, 14, 15, 17, 18, 26, 27, 28, 29, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 83, 84, 85, 86, 90, 94, 95, 98, 99, 100, 101, 102, 103, 104, 105, 108, 109, 110, 111, 112, 116, 117, 118, 121, 122, 123, 125, 126, 127, 140, 141, 142, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 162, 163, 164, 165, 169, 176, 177, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 203, 205, 207, 208, 217, 221, 225, 229, 232, 233, 235, 237, 238, 240, 241, 242, 244, 245, 246, 247, 248, 250, 251, 252, 256, 257, 259, 273, 274, 275, 278, 279, 280, 284, 286, 287, 291, 293, 294, 295, 296, 297, 298, 299, 300, 303, 305, 307, 310, 311, 312, 313, 326, 327, 328, 330, 331, 332, 333, 334, 335, 343, 349, 352, 355, 356, 358, 362, 363, 365, 370, 375, 378, 380, 381, 383, 386, 391, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 455, 456, 457, 458, 459, 460, 461, 463, 466, 467, 468, 476, 479, 486, 488], "ord": 189, "order": [0, 1, 28, 83, 101, 131, 186, 187, 189, 233, 237, 293, 324, 344, 376, 388, 463, 479, 481], "ordinari": 171, "org": [344, 348, 350, 357, 382, 413, 435], "origin": [5, 118, 309, 328, 371, 402, 403, 404, 405, 455, 456, 457, 458, 460, 461, 484], "orthonorm": 168, "ostream": 2, "ostringstream": 2, "other": [0, 2, 5, 7, 178, 189, 324, 364, 452, 461, 479, 480, 482, 483, 485], "other_input": 324, "otherwis": [18, 101, 123, 217, 247, 307, 310, 311, 312, 313, 363, 365, 375, 397, 399, 400, 415, 421, 426, 433, 445, 446, 483, 484], "our": [1, 2, 5, 6, 388, 455, 456, 457, 458, 460, 461, 480], "out": [0, 1, 2, 8, 90, 143, 326, 327, 337, 338, 355, 356, 372, 479, 480, 481, 482], "out_ax": [303, 481], "out_channel": [330, 331, 332, 333, 334, 335], "out_dim": [324, 452], "out_dtyp": 2, "out_idx": 2, "out_mask": 90, "out_proj": [5, 452], "out_ptr": 2, "out_shap": [1, 2], "outer": [0, 479, 483], "outlier": 429, "output": [0, 1, 2, 5, 8, 15, 16, 17, 18, 28, 83, 90, 91, 94, 96, 97, 108, 109, 110, 111, 112, 130, 140, 142, 143, 144, 145, 146, 153, 156, 157, 158, 162, 163, 165, 168, 169, 172, 189, 193, 203, 205, 207, 208, 221, 225, 229, 230, 233, 235, 236, 240, 241, 242, 244, 245, 246, 248, 251, 252, 265, 266, 273, 278, 280, 284, 288, 294, 296, 299, 300, 301, 302, 303, 304, 305, 306, 326, 327, 328, 330, 331, 332, 333, 334, 335, 348, 352, 355, 356, 378, 381, 397, 399, 400, 402, 403, 404, 405, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 446, 449, 479, 480, 481, 482, 483, 484, 485, 486, 487], "output_dim": [6, 324, 352, 381], "output_directori": 2, "output_dtyp": [1, 143], "output_fil": 5, "output_nam": [1, 143], "output_shap": [1, 143], "outsid": [143, 159], "over": [0, 2, 5, 6, 15, 17, 26, 27, 28, 29, 98, 99, 100, 101, 102, 103, 104, 108, 109, 110, 111, 149, 152, 155, 158, 171, 189, 191, 193, 203, 205, 207, 221, 233, 235, 261, 273, 274, 280, 284, 291, 293, 300, 328, 330, 331, 332, 333, 334, 335, 344, 350, 382, 423, 468, 471, 480, 481], "overal": 2, "overhead": [479, 483, 487], "overlap": 1, "overload": 18, "overrid": [2, 132], "overview": 3, "overwrit": 5, "own": [8, 484], "owndata": 484, "p": [8, 240, 324, 336, 337, 338, 434, 458, 460], "pack": [164, 237, 238], "packag": [2, 4, 6, 8, 319, 449, 480], "package_data": 2, "pad": [0, 1, 98, 99, 100, 101, 102, 103, 104, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 326, 327, 330, 331, 332, 333, 334, 335, 355, 356], "pad_valu": 0, "pad_width": [0, 232], "padding_hi": 0, "padding_lo": 0, "page": 485, "pain": 5, "pair": [0, 2, 232, 365, 386], "pairwis": 434, "pan": 5, "paper": [328, 391, 455, 456, 457, 458, 460, 461], "parallel": [480, 487], "param": [299, 324, 449, 481], "paramet": [0, 1, 2, 4, 5, 6, 12, 13, 14, 15, 16, 17, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 37, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 115, 116, 117, 118, 121, 122, 123, 125, 126, 127, 128, 129, 130, 131, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 216, 217, 218, 219, 221, 222, 223, 224, 225, 226, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 307, 308, 309, 310, 311, 312, 313, 314, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 348, 349, 350, 351, 352, 355, 356, 358, 359, 362, 363, 365, 370, 371, 372, 375, 376, 377, 378, 379, 380, 381, 382, 383, 386, 388, 391, 395, 397, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 414, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 446, 448, 449, 452, 454, 455, 456, 457, 458, 459, 460, 461, 462, 463, 465, 466, 467, 468, 469, 470, 471, 472, 474, 479, 480, 481, 483], "parameter_scal": 456, "parametr": [379, 436], "pars": 5, "parse_arg": 5, "parser": 5, "part": [1, 2, 170, 253, 481, 482], "partial": [376, 377, 479, 483], "particip": [121, 122, 125, 126, 127], "particular": [237, 344], "particularli": 479, "partit": [0, 28], "pass": [1, 2, 5, 6, 65, 79, 231, 232, 299, 308, 310, 311, 312, 324, 363, 375, 376, 377, 388, 479, 480, 483], "password": 480, "path": [3, 8, 131, 219, 265, 266, 307, 312, 365, 480], "pattern": [324, 483], "peak": [213, 215], "penalti": 467, "pep": 484, "per": [5, 6, 116, 141, 164, 237, 238, 307, 328, 344, 348, 350, 382, 474, 479, 480, 483], "perceptron": 7, "perf_count": 479, "perfectli": 483, "perform": [0, 1, 2, 3, 5, 7, 14, 90, 101, 108, 109, 110, 111, 127, 130, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 163, 164, 168, 186, 187, 204, 238, 260, 273, 287, 324, 344, 399, 404, 405, 454, 479, 480, 482, 483, 487], "perhap": [2, 5], "perm": 6, "permtuat": 247, "permut": [0, 6], "persist": 8, "pg": 189, "phi": [341, 411], "physic": 480, "pi": [134, 341, 391, 412, 481], "pick": 2, "pip": [2, 8], "pipelin": 2, "pixel": 337, "place": [2, 5, 38, 259, 260, 307, 480, 483, 484], "placehold": 479, "plai": [2, 5], "plain": 388, "plan": [2, 479], "platform": 8, "plu": [0, 197], "point": [0, 2, 4, 5, 8, 83, 161, 238, 317], "pointer": 2, "pool": [326, 327, 355, 356, 487], "popul": 2, "portion": 336, "posinf": [0, 225], "posit": [0, 5, 28, 118, 145, 159, 165, 177, 183, 184, 223, 225, 233, 245, 259, 294, 299, 311, 324, 330, 331, 332, 333, 334, 335, 378, 386, 391, 424, 434], "possibl": [275, 340, 380, 479, 480, 482, 487], "possibli": [5, 14, 90, 163, 204, 309], "postur": 5, "potenti": 217, "power": [0, 481, 484], "practic": [2, 479], "pre": [8, 146, 421], "preced": 344, "precis": [0, 2, 5, 139, 146, 324, 341, 382, 421, 462, 479], "preclud": 324, "pred": [425, 429], "predic": [307, 370], "predict": [421, 424, 425, 426, 427, 428, 429, 431, 432, 433], "prefix": [303, 310], "prelu": 324, "prepar": [2, 5], "prepend": [3, 204], "preprint": [5, 455, 461], "preprocessor": 8, "present": 1, "preserv": [257, 481], "press": [5, 189], "pressur": 2, "pretti": [479, 483], "prevent": [281, 434, 484], "previou": [216, 217, 218], "primal": [1, 2, 112, 179, 302], "primit": 481, "print": [1, 2, 4, 5, 6, 8, 309, 310, 311, 312, 314, 324, 476, 479, 480, 481, 482, 483, 484, 485], "prior": [236, 287, 288], "priorit": 481, "privat": 2, "prng": [240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 476], "prob": 421, "probabl": [8, 248, 336, 337, 338, 381, 421, 423, 427, 487], "problem": [4, 6, 324], "process": [5, 101, 105, 120, 121, 122, 123, 125, 126, 127, 311, 312, 338, 340, 399, 478, 480], "processor": 8, "prod": [0, 1], "produc": [0, 2, 8, 94, 378, 449], "product": [0, 2, 14, 83, 110, 171, 179, 185, 204, 231, 235, 291, 302, 378, 485], "profil": 3, "program": [213, 480], "programmat": 377, "project": [3, 5, 378], "project_source_dir": 2, "promot": 2, "promote_typ": 2, "promoted_dtyp": 2, "prompt": 5, "propag": [481, 482], "properti": [31, 38, 47, 51, 61, 62, 68, 70, 371, 374, 464, 481], "proportion": 309, "protocol": 484, "provid": [0, 2, 5, 83, 116, 141, 165, 247, 259, 291, 299, 311, 313, 319, 324, 358, 363, 365, 375, 376, 377, 380, 381, 399, 400, 448, 452, 480, 486, 488], "pseudo": 476, "pth": 5, "public": [2, 324], "pun": 0, "pure": [1, 324, 454], "purpos": [1, 189], "purs": 5, "push": 2, "push_back": 2, "put": [0, 1, 6, 236, 479, 480], "put_along_axi": 0, "py": [2, 5, 8, 480], "pypi": 8, "python": [1, 3, 5, 50, 68, 78, 136, 310, 311, 312, 313, 314, 452, 462, 463, 465, 478, 480, 481, 484], "python_requir": 2, "pytorch": [5, 7, 341, 344, 481], "pytorch_compat": 344, "q": [146, 190], "qualifi": 480, "quantiz": [0, 116, 141, 164, 194, 238, 380, 381], "quantized_matmul": 0, "quantizedembed": 324, "quantizedlinear": 324, "quarter": 5, "queri": [5, 146, 218, 378], "query_input_dim": 378, "query_proj": 5, "question": [5, 483], "queue": 3, "quick": [2, 7], "quit": [481, 484], "quotient": [0, 128, 129, 161], "r": [2, 5, 190, 299, 337, 343], "r_t": 343, "race": 487, "radian": [0, 115], "rag": 5, "rain": 5, "rais": [0, 5, 189, 217, 234, 275, 365], "ram": 5, "random": [1, 2, 3, 4, 5, 6, 7, 143, 326, 327, 328, 348, 355, 356, 365, 372, 479, 481, 487, 488], "randomli": [4, 5, 247, 336, 337, 338], "rang": [0, 2, 3, 4, 5, 6, 8, 18, 159, 163, 193, 403, 405, 412, 413, 454, 468, 469, 470, 471, 472, 476, 479, 481, 483, 487], "rank": [0, 125, 126, 127, 430, 480], "rate": [4, 454, 455, 456, 457, 458, 459, 460, 461, 466, 467], "rather": [2, 481, 487], "ratio": [0, 24], "rceil": 90, "re": [6, 8, 449], "readabl": 3, "readi": 2, "real": [0, 153, 154, 155, 156, 157, 158, 183, 184, 186, 187], "realli": 350, "reason": [1, 5, 482], "reboot": 8, "receiv": [125, 126, 307, 470, 484], "reciproc": [0, 261], "reclaim": 216, "recommend": [8, 217, 461], "recompil": [94, 479], "record": [3, 213, 483], "recreat": [314, 454], "rectifi": [351, 384, 385, 404, 405, 418, 437, 438], "recurr": [343, 349, 383], "recurs": [324, 362, 363, 368, 373, 375, 452], "recv": 126, "redirect": 2, "reduc": [0, 1, 8, 15, 17, 26, 27, 122, 203, 205, 207, 221, 235, 280, 284, 300, 313, 328, 399, 429], "reduct": [15, 17, 122, 203, 205, 221, 235, 313, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 480], "redund": 481, "refer": [189, 348, 357, 371, 402, 403, 404, 405, 413, 435, 482], "reflect": [371, 479, 482, 484], "regard": 341, "regardless": [83, 146], "regist": [2, 6], "register_librari": 2, "regress": [7, 429], "regular": [38, 337, 435, 459, 479, 482], "regularli": 2, "reimplement": 2, "rel": [16, 172, 456, 479], "relative_step": 456, "relax": 217, "relev": 2, "reli": [1, 2], "relu": [324, 379, 399, 436, 449], "relu6": 324, "remain": [0, 5, 218, 299, 312, 336, 337, 338, 480], "remaind": [0, 129], "remov": [0, 118, 204, 241, 278, 423], "rep": [0, 292], "repeat": [0, 292], "repeatedli": 4, "repetit": 256, "replac": [0, 5, 225, 376, 377, 399, 433], "replai": 3, "repli": 5, "repo": [4, 6, 8, 479], "report": [211, 217], "repres": [2, 5, 120, 123, 164, 430, 434, 484], "represent": [5, 237, 301, 310, 314], "request": 2, "requir": [1, 2, 5, 324, 480, 483, 484], "requires_grad": 481, "rerun": [479, 483], "rescal": 309, "research": 7, "reset": 215, "reset_peak_memori": 213, "reshap": [0, 5, 189, 400, 482], "resid": 218, "resolv": 2, "resourc": 2, "respect": [2, 4, 6, 142, 144, 163, 164, 165, 237, 299, 311, 324, 328, 341, 344, 348, 350, 452, 481, 485], "respons": 2, "rest": [5, 145, 311, 312, 386], "restart": 8, "restor": 259, "result": [0, 5, 14, 18, 38, 78, 83, 94, 142, 144, 164, 189, 204, 238, 245, 256, 279, 311, 312, 313, 391, 421, 479, 481, 484], "resum": 5, "return": [0, 1, 2, 4, 5, 6, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 37, 50, 68, 78, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 115, 116, 117, 118, 121, 122, 123, 125, 126, 127, 128, 129, 130, 131, 133, 134, 135, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 212, 216, 217, 218, 221, 222, 223, 224, 225, 226, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 260, 261, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 307, 308, 309, 310, 311, 312, 313, 314, 324, 343, 349, 358, 359, 360, 362, 363, 364, 365, 366, 367, 368, 372, 373, 375, 376, 377, 383, 401, 402, 403, 404, 405, 406, 407, 408, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 449, 452, 462, 478, 479, 480, 481, 482, 483, 484, 486, 487], "return_metadata": 194, "revers": [0, 2, 41, 42, 43, 44, 83, 108, 109, 110, 111, 295, 391], "rf": 8, "rfft": 153, "rfft2": 154, "rfftn": 155, "rfloor": [326, 327, 355, 356], "rho": 455, "rhs_indic": [0, 163, 164], "rhs_mask": 90, "right": [0, 1, 2, 8, 237, 258, 259, 326, 327, 341, 355, 356, 400, 412, 413, 424, 426, 434], "right_shift": 0, "rm": [5, 8, 144, 456], "rmsnorm": [5, 324], "rmsprop": 454, "rnn": [324, 343], "roadcast": 248, "robust": 429, "roform": [5, 386], "roll": 0, "root": [0, 5, 144, 261, 276, 382], "rope": [5, 324], "rosetta": 8, "rotari": [5, 145, 386], "rotat": [145, 386], "round": [0, 237], "routin": 2, "row": [0, 1, 2, 83, 140, 143, 169, 237, 296], "row_contigu": 2, "rpath": 2, "rsqrt": 0, "rtol": [0, 16, 172], "rule": [2, 454], "run": [1, 2, 3, 5, 6, 7, 8, 9, 143, 231, 315, 328, 358, 455, 456, 458, 459, 460, 479, 480, 483, 487, 488], "runtim": [5, 123, 319, 479, 480], "runtime_error": 2, "safetensor": [8, 194, 264, 365, 369, 454, 483, 486], "sai": [2, 5, 449, 483], "said": 5, "sake": 481, "same": [0, 2, 5, 8, 16, 38, 82, 91, 94, 99, 100, 101, 103, 104, 105, 121, 142, 144, 153, 156, 157, 158, 164, 165, 172, 179, 232, 241, 259, 260, 301, 302, 304, 312, 324, 327, 328, 336, 344, 348, 356, 380, 401, 402, 403, 404, 405, 406, 407, 408, 423, 434, 452, 462, 476, 479, 480, 482, 487], "sampl": [2, 4, 5, 193, 240, 241, 242, 244, 245, 248, 251, 252, 402, 403, 404, 405, 407, 408, 424, 430, 434, 476, 479], "sat": 5, "save": [3, 5, 7, 194, 219, 237, 263, 264, 265, 266, 369, 483], "save_gguf": 486, "save_safetensor": [369, 454, 486], "save_weight": 324, "savez": [5, 369, 486], "savez_compress": 486, "saw": [5, 481], "scalar": [0, 2, 13, 14, 16, 30, 50, 78, 82, 87, 88, 89, 90, 91, 93, 128, 129, 133, 161, 162, 165, 166, 167, 168, 172, 180, 181, 182, 193, 199, 200, 201, 202, 204, 206, 222, 224, 225, 228, 232, 234, 240, 248, 251, 252, 255, 258, 263, 283, 299, 301, 304, 308, 434, 481, 483, 485], "scale": [0, 2, 5, 14, 116, 141, 142, 144, 145, 146, 164, 168, 237, 238, 244, 246, 309, 337, 338, 350, 378, 386, 387, 391, 400, 439, 456], "scale_arr": 2, "scale_factor": 400, "scale_paramet": 456, "scatter": 0, "scatter_add": 0, "scatter_max": 0, "scatter_min": 0, "scatter_prod": 0, "schedul": [2, 217, 454, 468, 469, 470, 471, 472, 474, 487], "schema": 3, "scipi": 168, "scope": 324, "score": [5, 146, 430], "sdk": 8, "se": 1, "second": [5, 8, 118, 178, 180, 200, 202, 204, 258, 285, 294, 299, 327, 356, 422, 430, 456, 458, 459, 460, 481, 487], "second_layer_a": 483, "second_layer_b": 483, "secret": 5, "section": [1, 5, 8, 275, 434, 479, 480, 481], "see": [1, 2, 5, 6, 8, 10, 11, 32, 33, 34, 35, 36, 39, 40, 41, 42, 43, 44, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 189, 216, 263, 264, 307, 317, 324, 328, 329, 337, 339, 341, 345, 346, 347, 353, 354, 361, 379, 380, 381, 384, 385, 386, 387, 389, 391, 392, 393, 394, 395, 396, 398, 400, 402, 403, 404, 405, 411, 412, 413, 439, 479, 480, 481, 482, 485, 487], "seed": 243, "seen": 484, "select": [0, 3, 8, 186, 187, 293, 304, 358, 362, 370], "self": [5, 6, 9, 30, 31, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 50, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 112, 315, 324, 435, 452], "selu": 324, "semant": [13, 87, 88, 89, 91, 128, 129, 133, 166, 167, 180, 181, 182, 199, 204, 206, 222, 224, 228, 234, 255, 258, 283, 487], "semi": [183, 184, 245], "send": 480, "sennrich": 5, "sensit": 429, "sentencepiec": 5, "separ": [5, 65, 79, 344, 430], "sequenc": [5, 15, 17, 33, 34, 56, 57, 58, 59, 63, 71, 74, 75, 76, 80, 83, 91, 101, 125, 138, 143, 148, 149, 151, 152, 154, 155, 157, 158, 162, 203, 205, 207, 221, 229, 235, 240, 241, 242, 244, 245, 246, 248, 251, 252, 257, 273, 275, 278, 280, 284, 291, 292, 295, 300, 305, 328, 330, 333, 343, 349, 383, 399, 476, 487], "sequenti": [324, 449], "seri": 8, "serial": 454, "set": [2, 5, 6, 8, 94, 112, 119, 121, 122, 123, 125, 126, 127, 132, 142, 145, 210, 216, 217, 218, 267, 268, 282, 341, 350, 352, 361, 363, 370, 371, 372, 375, 376, 381, 386, 397, 422, 434, 446, 452, 454, 456, 463, 476, 481, 483], "set_data": 2, "set_default_devic": 2, "set_dtyp": 324, "set_input_arrai": 2, "set_memory_limit": 216, "set_output_arrai": 2, "setbyt": 2, "setcomputepipelinest": 2, "setup": [2, 4, 6, 8, 479], "sever": [5, 8, 98, 99, 100, 101, 102, 103, 104, 265, 266, 479, 486], "sgd": [4, 6, 454, 461, 463, 468, 469, 472, 479], "shade": [1, 2], "shall": 5, "shape": [0, 2, 3, 5, 6, 65, 82, 83, 90, 91, 94, 98, 99, 100, 101, 102, 103, 104, 118, 121, 125, 126, 143, 146, 147, 150, 153, 156, 157, 158, 162, 163, 168, 179, 188, 192, 204, 229, 230, 240, 241, 242, 244, 245, 246, 248, 251, 252, 257, 259, 301, 302, 304, 305, 306, 324, 326, 327, 328, 330, 331, 332, 333, 334, 335, 337, 338, 343, 348, 349, 352, 355, 356, 365, 383, 401, 402, 403, 404, 405, 406, 407, 408, 423, 434, 454, 479, 481, 482, 485, 487], "shapeless": [0, 94], "share": [7, 116, 141, 164, 237, 238, 301, 480], "shazeer": 5, "shift": [0, 180, 258, 259, 328], "shop": 5, "should": [1, 2, 4, 5, 6, 8, 83, 118, 121, 142, 143, 144, 146, 179, 209, 218, 219, 236, 237, 288, 294, 299, 302, 307, 310, 324, 330, 331, 332, 333, 334, 335, 337, 338, 372, 378, 388, 423, 425, 430, 452, 478, 479, 480, 481, 483, 484, 488], "show": [8, 317, 479], "shown": 2, "shuffl": 6, "side": [0, 232, 326, 327, 355, 356, 479], "sigma": [341, 342, 343, 349, 390, 402, 403, 404, 405, 413, 414, 419, 440, 441], "sigmoid": [0, 5, 324, 353, 389, 413, 419, 421, 441], "sign": [0, 16, 172, 317, 461], "signal": [105, 400], "signatur": [1, 143], "signedinteg": [11, 178], "signific": 237, "silent": [156, 157, 158], "silicon": [2, 5, 7, 8, 487], "silu": 324, "simd": 1, "simd_sum": 1, "simdgroup": 1, "simdgroup_s": 1, "similar": [5, 164, 178, 311, 376, 377, 378, 422, 484, 486], "similarli": [2, 8, 204, 481, 483], "simpl": [2, 5, 6, 324, 340, 448, 454, 479, 480, 481, 483], "simple_axpbi": 2, "simple_tim": 2, "simplest": [2, 324, 480], "simpli": [2, 5, 8, 339, 351, 384, 410, 418, 437, 447, 452, 479, 480, 481], "simplic": 0, "simultan": 1, "sin": [0, 112, 391, 481, 485], "sinc": [1, 2, 5, 6, 164, 213, 452, 461, 470, 484, 487], "sine": [0, 21, 22, 271, 272, 481], "sing": 189, "singer": 457, "singl": [2, 6, 136, 179, 194, 208, 232, 302, 327, 356, 479, 482, 486], "singleton": [0, 15, 17, 26, 27, 123, 203, 204, 205, 207, 221, 235, 280, 284, 300], "singular": [189, 191], "sinh": 0, "sinusoid": 391, "sinusoidalpositionalencod": 324, "size": [0, 1, 2, 5, 6, 51, 68, 90, 99, 100, 103, 104, 116, 138, 141, 142, 143, 144, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 162, 164, 168, 169, 178, 185, 189, 212, 217, 218, 237, 238, 241, 257, 275, 278, 301, 307, 324, 326, 327, 330, 331, 332, 333, 334, 335, 340, 348, 355, 356, 380, 381, 400, 456, 480, 483, 484], "size_in_megabyt": 218, "size_t": [0, 2], "sizeof": 2, "skip": [3, 83], "slice": [0, 482], "slice_s": 0, "slice_upd": 0, "slight": [5, 483], "slightli": [386, 487], "slope": 351, "slot": 480, "slow": 479, "slowli": 5, "small": [5, 139, 142, 144, 328, 344, 350, 382, 424, 429, 434, 479, 480, 487], "smaller": [0, 8, 233, 461, 479], "smallest": 189, "smile": 5, "smooth": [423, 433, 466], "smooth_l1_loss": 324, "sned": 127, "snippet": 480, "so": [1, 2, 5, 8, 165, 168, 299, 336, 400, 454, 479, 480, 483, 487], "softmax": [0, 5, 146, 324, 354, 420, 423], "softmin": 324, "softplu": [324, 357, 435], "softshrink": 324, "softsign": 324, "solv": 324, "some": [0, 2, 4, 5, 6, 363, 375, 454, 463, 479, 481, 483], "someon": 5, "someth": [4, 5, 482], "sonoma": 8, "soon": 5, "sort": [0, 28, 29, 233, 293], "sourc": [0, 1, 2, 3, 60, 125, 126, 143, 223, 295, 480], "space": [0, 2, 193, 421, 432], "spars": [0, 208], "spatial": [99, 100, 101, 103, 104, 326, 344, 355, 400], "speak": [5, 189], "special": 2, "specif": [1, 2, 8, 480, 481], "specifi": [0, 2, 18, 37, 99, 100, 101, 103, 104, 118, 154, 155, 162, 165, 185, 189, 193, 223, 229, 236, 241, 256, 285, 287, 288, 291, 294, 295, 299, 303, 305, 328, 397, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 446, 480, 481, 487], "speed": [1, 2], "spent": 5, "split": [0, 342, 344, 414], "splittabl": 476, "sqrt": [0, 5, 134, 146, 168, 328, 341, 344, 348, 350, 352, 382, 391, 402, 403, 404, 405, 412, 455, 457, 458, 459, 466, 479], "squar": [0, 4, 5, 144, 169, 188, 192, 261, 276, 299, 311, 324, 382, 431, 433, 455, 456, 458, 459, 460, 481, 484], "squeez": [0, 400, 479], "src": [0, 125, 126], "ssh": 480, "stabil": [142, 144, 328, 344, 348, 350, 382, 421, 422, 424, 455, 456, 457, 458, 459, 460, 466], "stabl": [199, 203, 273, 429], "stable_abi": 2, "stack": [0, 479], "standard": [0, 1, 6, 50, 78, 204, 242, 246, 280, 399, 402, 404, 407, 480, 485], "starmap": [5, 311], "start": [0, 1, 2, 4, 5, 7, 8, 18, 145, 193, 219, 275, 313, 479, 482, 487], "start_axi": [0, 49, 159], "start_captur": 3, "state": [5, 6, 324, 343, 349, 383, 454, 463, 476, 479], "static": 8, "static_cast": 2, "std": [0, 2, 407], "step": [0, 3, 5, 6, 18, 324, 343, 349, 383, 456, 463, 468, 470, 471, 472, 479, 480], "step_decai": 454, "step_siz": 472, "still": [5, 8, 189, 479, 483], "stochast": [457, 458, 460, 467, 483], "stood": 5, "stop": [0, 2, 5, 18, 193, 220, 281, 481, 482], "stop_captur": 3, "stop_gradi": [0, 481], "storag": 83, "store": 5, "str": [2, 105, 130, 131, 143, 165, 186, 187, 189, 194, 208, 210, 219, 262, 263, 264, 265, 266, 299, 310, 314, 358, 359, 362, 363, 365, 367, 369, 375, 400, 404, 405, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434], "straight": 5, "strang": 5, "stream": [2, 7, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 114, 115, 116, 117, 118, 121, 122, 125, 126, 127, 128, 129, 130, 133, 134, 135, 137, 138, 139, 140, 141, 142, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 300, 301, 304, 305, 306, 480, 487], "streamcontext": 282, "streamordevic": [0, 2], "street": 5, "strength": [461, 467], "strict": [123, 166, 181, 363, 365, 375], "strictli": [189, 218], "stride": [0, 2, 83, 98, 99, 100, 101, 102, 103, 104, 326, 327, 330, 331, 332, 333, 334, 335, 355, 356, 386, 482], "string": [0, 2, 131, 143, 210, 232, 484, 486], "structur": [2, 462, 481], "stub": 8, "style": [2, 13, 16, 87, 88, 89, 128, 129, 133, 166, 167, 172, 180, 181, 182, 199, 204, 206, 222, 224, 228, 234, 255, 258, 283], "su": 5, "sub": [0, 6, 118, 250, 294, 307], "subarrai": [118, 275], "subclass": 452, "subdivid": 1, "subdtyp": 178, "subgradi": 457, "sublinear": 456, "submodul": [5, 6, 324, 359, 363, 364, 375, 377], "subscript": [130, 131], "subsect": 5, "subsequ": 454, "subset": [324, 362], "substanti": 8, "subtract": [0, 38], "subtyp": [178, 317], "sudo": [8, 218], "sum": [0, 2, 4, 13, 111, 122, 171, 189, 203, 273, 291, 294, 324, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 480, 482, 484], "sum_": [189, 326, 327, 429], "sum_i": 420, "sum_j": [442, 443], "summat": [130, 131], "super": [5, 6, 324, 452], "superset": [311, 462], "support": [1, 2, 5, 7, 8, 16, 90, 100, 103, 104, 146, 159, 168, 172, 183, 184, 186, 187, 188, 190, 191, 192, 194, 204, 237, 245, 480, 481, 482, 484, 486], "suppos": [481, 487], "sure": [2, 3, 5, 8, 324, 479], "surpass": [404, 405], "surpris": 5, "sw": 1, "swap": [0, 105, 217, 285, 377], "swapax": [0, 112], "swiglu": 5, "swish": [389, 441], "switch": 8, "symbol": 461, "symmetr": [99, 100, 103, 104, 183, 184, 186, 187], "symmetri": [186, 187], "synchron": [2, 479], "syntax": [38, 482], "synthet": 4, "sysctl": 218, "system": [5, 8, 210, 211, 212, 218], "t": [0, 1, 2, 5, 8, 134, 143, 146, 164, 183, 184, 238, 299, 324, 326, 343, 349, 355, 383, 455, 456, 457, 458, 459, 460, 461, 466, 467, 479, 481, 487], "tabl": [1, 189, 317, 340], "take": [0, 2, 5, 6, 87, 88, 89, 94, 163, 165, 179, 206, 222, 230, 238, 288, 299, 302, 303, 306, 312, 313, 378, 421, 476, 480, 481, 482, 486, 487, 488], "take_along_axi": [0, 482], "taken": [118, 287, 294], "talk": 480, "tan": 0, "tangent": [0, 2, 23, 24, 25, 112, 179, 289, 290, 398, 447], "tangent_i": 2, "tangent_x": 2, "tanh": [0, 324, 341, 343, 349, 357, 383, 412, 435], "target": [2, 299, 421, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 479], "target_include_directori": 2, "target_link_librari": 2, "target_link_opt": 2, "target_sourc": 2, "task": [217, 429], "tau": 467, "tcp": 480, "tell": [5, 479, 484], "temp": 5, "templat": [0, 1, 2, 143], "ten": [481, 483], "tend": 461, "tensor": [194, 291, 326, 327, 355, 356, 434, 484], "tensordot": 0, "term": [2, 424, 455, 456, 457, 458, 459, 460, 466], "termin": 8, "test": [6, 8, 480], "test_imag": 6, "test_label": 6, "text": [5, 326, 327, 341, 343, 349, 355, 356, 357, 383, 390, 397, 402, 403, 404, 405, 412, 415, 416, 417, 424, 425, 426, 429, 430, 433, 435, 436, 439, 440, 445, 446, 456, 461], "textrm": [237, 341, 342, 411, 414], "tf": 484, "tgp_size": 2, "th": [108, 109, 110, 111, 117, 140, 186, 470], "than": [1, 2, 5, 78, 105, 118, 129, 145, 163, 166, 167, 181, 182, 183, 184, 186, 187, 188, 191, 192, 204, 216, 218, 309, 311, 386, 397, 400, 430, 433, 446, 456, 461, 479, 481, 487], "thank": 483, "thei": [1, 2, 4, 5, 8, 16, 105, 164, 172, 388, 425, 452, 461, 478, 479, 480, 483, 485, 486, 487], "them": [0, 2, 5, 121, 324, 363, 375, 480, 487], "themselv": [2, 479], "thi": [0, 1, 2, 5, 6, 8, 15, 16, 17, 18, 26, 27, 28, 29, 83, 112, 132, 143, 163, 164, 168, 172, 179, 183, 184, 186, 187, 188, 189, 190, 191, 192, 199, 203, 204, 205, 207, 209, 211, 218, 221, 233, 235, 241, 268, 273, 274, 275, 280, 284, 287, 293, 300, 309, 312, 313, 324, 336, 337, 338, 342, 343, 349, 359, 360, 362, 363, 366, 367, 368, 373, 375, 376, 377, 378, 381, 383, 397, 402, 403, 404, 405, 412, 413, 414, 421, 429, 446, 452, 463, 478, 479, 480, 481, 483, 484, 486], "thing": [2, 5, 480], "third": 185, "thompson": 337, "those": [2, 5, 324], "though": [2, 5, 479, 483, 484], "thousand": 483, "thread": [1, 2], "thread_index_in_simdgroup": 1, "thread_position_in_grid": [1, 2, 143], "threadgroup": [1, 2, 143], "threads_per_simdgroup": 1, "three": [5, 86, 400], "threefri": 476, "threshold": [397, 426, 433, 446], "through": [1, 2, 281, 399, 461, 479, 481, 484], "throw": [2, 94, 123], "thu": [5, 324], "thumb": 454, "tic": 479, "tieleman": 466, "tile": [0, 146], "time": [2, 5, 8, 217, 292, 324, 326, 327, 343, 349, 355, 356, 383, 479, 481, 483, 487], "timeit": [479, 481], "titl": 2, "tmp": [1, 143], "to_quant": 307, "to_stream": 2, "toc": 479, "togeth": [0, 1, 2, 6, 237, 311, 312, 480], "tok_embed": 5, "token": [5, 340, 380], "told": 5, "toler": [0, 16, 172], "too": [178, 479, 483], "took": 5, "tool": 8, "top": [2, 293, 352, 400], "topk": 0, "torch": [5, 484], "torch_weight": 5, "total": [218, 481], "total_norm": 309, "tpi": 479, "trace": [0, 3, 479], "trace_fil": 3, "tracer": 376, "track": [2, 324, 328], "track_running_stat": 328, "trade": 483, "tradit": [5, 145, 337, 338, 386], "train": [5, 6, 324, 328, 336, 337, 338, 361, 363, 375, 402, 403], "train_imag": [6, 454], "train_label": [6, 454], "trainabl": [6, 308, 324, 452], "trainable_paramet": [324, 362, 463], "transform": [1, 5, 7, 112, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 168, 308, 324, 328, 344, 350, 352, 362, 363, 375, 381, 386, 482], "transformerencod": 265, "transit": 470, "translat": [142, 350], "transpos": [0, 5, 31, 102, 103, 104, 164, 238, 333, 334, 335], "treat": [0, 2, 154, 155, 157, 158, 287, 400, 479], "tree": [7, 94, 136, 165, 299, 303, 310, 311, 312, 313, 314, 462, 463, 465, 474, 481], "tree_flatten": [265, 311, 314, 324, 454], "tree_map": [312, 324, 480], "tree_unflatten": [5, 454], "trembl": 5, "tri": 0, "triangl": [186, 187, 296], "triangular": [183, 184, 192], "tril": 0, "trilinear": 400, "triplet": 434, "triplet_loss": 324, "triu": 0, "true": [0, 1, 2, 4, 5, 16, 41, 42, 43, 44, 82, 94, 108, 109, 110, 111, 143, 145, 164, 172, 178, 183, 184, 189, 194, 208, 217, 238, 273, 304, 307, 310, 311, 312, 313, 317, 324, 328, 330, 331, 332, 333, 334, 335, 343, 344, 348, 349, 350, 352, 362, 363, 365, 372, 375, 381, 383, 386, 391, 399, 400, 421, 429, 456], "truncat": [147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 251], "truth": [4, 423, 433], "try": [2, 8], "tupl": [0, 30, 65, 68, 79, 95, 99, 100, 101, 103, 104, 125, 129, 131, 136, 138, 179, 186, 189, 190, 191, 232, 237, 257, 259, 278, 299, 302, 310, 311, 312, 313, 314, 326, 327, 331, 332, 334, 335, 355, 356, 365, 367, 388, 400, 456, 458, 459, 460, 461, 478, 481], "tutori": 2, "twice": 487, "two": [0, 2, 13, 14, 16, 24, 82, 85, 87, 88, 89, 90, 118, 128, 133, 148, 151, 157, 163, 164, 166, 167, 172, 181, 182, 183, 184, 185, 186, 187, 188, 190, 191, 192, 199, 204, 206, 222, 224, 228, 231, 285, 313, 327, 342, 349, 356, 414, 422, 479, 480, 481, 482, 487], "txt": 2, "type": [0, 1, 2, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 37, 68, 78, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 115, 116, 117, 118, 121, 122, 123, 125, 126, 127, 128, 129, 130, 131, 133, 134, 135, 137, 138, 139, 140, 141, 142, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 210, 216, 217, 218, 221, 222, 223, 224, 225, 226, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 250, 251, 252, 253, 254, 255, 256, 257, 258, 260, 261, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 283, 284, 285, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 309, 310, 313, 324, 370, 399, 401, 402, 403, 404, 405, 406, 407, 408, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 479, 482], "type_nam": 2, "type_to_nam": 2, "typenam": [0, 1, 2], "typic": [0, 146, 340, 454, 479, 483], "u": [1, 2, 183, 186, 187, 191, 352, 377, 474, 483], "u_": 455, "u_t": 455, "uint": [1, 2, 143], "uint16": [11, 317], "uint3": 1, "uint32": [11, 26, 27, 28, 29, 241, 317], "uint64": [11, 317], "uint8": [11, 317], "ultra": 5, "unabl": 8, "unam": 8, "unari": 479, "unchang": [145, 281, 386], "uncheck": 8, "uncompress": 265, "undefin": [0, 28, 112, 183, 184, 233, 245, 482], "under": [2, 189], "underli": [2, 301], "understand": [5, 402, 403], "unexpect": [2, 18], "unfreez": [324, 363], "unfrozen": 375, "unifi": 7, "uniform": [3, 324, 352, 365, 403, 405, 449, 476, 479, 481, 487], "uniformli": 252, "unintend": 0, "union": [18, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 48, 49, 52, 53, 54, 55, 56, 57, 58, 59, 60, 63, 64, 65, 66, 67, 69, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 84, 85, 86, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 176, 177, 178, 186, 187, 210, 263, 282], "uniqu": [2, 476], "unique_ptr": 2, "unit": [329, 339, 341, 342, 343, 351, 384, 385, 387, 389, 402, 403, 404, 405, 409, 410, 411, 412, 413, 414, 418, 437, 438, 439, 441], "unittest": 8, "univers": 189, "unless": [5, 16, 172, 189, 452], "unlik": [5, 16, 172, 337, 338, 371], "unnecessari": [2, 5], "unnorm": [241, 421, 423], "unscal": 456, "unsign": [164, 237, 238, 317], "unsignedinteg": 11, "unspecifi": [15, 17, 18, 26, 27, 28, 29, 95, 108, 109, 110, 111, 162, 203, 205, 207, 221, 229, 233, 235, 256, 273, 274, 280, 284, 287, 293, 294, 300, 305, 488], "unsqueez": 5, "unsupport": 194, "until": [2, 483, 485], "unus": 2, "up": [1, 2, 5, 112, 479], "upcast": 2, "updat": [0, 1, 2, 4, 5, 6, 38, 94, 307, 311, 313, 328, 358, 359, 365, 370, 371, 372, 377, 454, 456, 459, 461, 462, 463, 467, 468, 469, 470, 471, 472, 479, 480, 483], "update_modul": 324, "uplo": [186, 187], "upon": [5, 311, 312], "upper": [183, 184, 186, 187, 192, 237, 248, 251, 252, 408], "upsampl": 324, "us": [0, 3, 4, 5, 6, 7, 8, 18, 38, 83, 112, 116, 119, 121, 122, 125, 126, 127, 129, 141, 143, 145, 159, 164, 180, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 204, 211, 212, 213, 216, 218, 237, 238, 256, 257, 258, 259, 286, 310, 313, 317, 319, 324, 327, 337, 340, 341, 343, 349, 352, 356, 358, 362, 369, 376, 378, 380, 381, 383, 386, 391, 399, 400, 404, 405, 412, 413, 422, 449, 452, 454, 455, 456, 458, 459, 460, 461, 462, 463, 476, 478, 479, 480, 481, 482, 485, 487], "usag": [112, 399, 479], "user": [2, 5, 324], "usual": [340, 380, 478, 483], "util": [1, 2, 5, 7, 8, 265, 324, 454, 480], "v": [5, 105, 146, 186, 324, 363, 484], "v_": [455, 457, 458, 459, 460, 466, 467], "v_t": [455, 457, 458, 459, 460, 466, 467], "val": [0, 30, 162], "valid": [6, 105, 159, 303, 310, 363, 375, 478], "valid_parameter_filt": 358, "valu": [0, 1, 4, 5, 11, 12, 16, 18, 26, 27, 50, 78, 82, 93, 140, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 162, 172, 185, 189, 191, 193, 210, 218, 225, 232, 236, 240, 241, 242, 244, 245, 246, 248, 251, 252, 259, 263, 287, 288, 299, 303, 308, 310, 311, 312, 313, 317, 327, 329, 336, 337, 338, 339, 345, 348, 352, 356, 362, 378, 379, 395, 397, 399, 401, 421, 422, 423, 424, 425, 426, 428, 429, 430, 431, 432, 433, 446, 452, 456, 459, 468, 469, 471, 472, 481], "value_and_grad": [6, 112, 324, 376, 452, 454, 465, 479, 481, 484, 485], "value_and_grad_fn": 483, "value_cach": 5, "value_dim": 378, "value_input_dim": 378, "value_output_dim": 378, "value_proj": 5, "valueerror": [189, 365, 481], "values_hat": 5, "van": 189, "var": [0, 328, 344, 348, 350, 424], "variabl": [8, 94, 119, 132, 165, 179, 299, 302, 303, 480], "varianc": [0, 280, 300, 328, 344, 424], "variant": [5, 433, 460], "variou": 189, "vector": [0, 2, 4, 7, 171, 179, 189, 287, 302, 303, 340, 423, 485], "verbos": [1, 143], "veri": [5, 378, 480, 483, 487], "verifi": [4, 8], "versa": 259, "version": [2, 8, 116, 141, 199, 203, 237, 273, 303, 476, 481, 482], "versu": 479, "via": [8, 112, 462, 465, 480, 483, 484], "vice": 259, "video": 338, "view": [0, 3, 83, 484], "virtual": 2, "vjp": [2, 112, 485], "vmap": [2, 112, 481, 483, 485], "vmap_add": 481, "vocab_s": 5, "vocabulari": [340, 380], "void": [1, 2], "vt": 191, "w": [0, 1, 4, 99, 100, 103, 104, 116, 141, 164, 186, 237, 238, 299, 312, 327, 328, 331, 332, 334, 335, 337, 338, 352, 356, 454, 467, 481], "w1": [5, 309], "w2": [5, 309], "w3": 5, "w_": [327, 343, 349, 356, 383, 455, 456, 457, 458, 459, 460, 461, 466, 467], "w_1": 237, "w_g": 237, "w_i": [116, 141, 237], "w_in": 1, "w_q": 237, "w_star": 4, "w_stride": 1, "w_t": [455, 457, 458, 459, 460, 461, 466, 467], "wa": [5, 83, 125, 126, 480, 483], "wai": [2, 5, 8, 324, 400, 479, 480, 481, 482], "wait": [2, 5, 217], "walk": 5, "walkthrough": 2, "walsh": 168, "want": [1, 5, 480, 481, 487], "warm": [2, 479], "warmup": [470, 471], "warmup_init": 456, "watch": [5, 479], "wd": 461, "we": [0, 1, 2, 4, 5, 6, 116, 125, 126, 141, 164, 237, 238, 324, 340, 380, 388, 459, 461, 476, 478, 479, 480, 481, 483, 487], "weight": [0, 4, 98, 99, 100, 101, 102, 103, 104, 142, 144, 311, 324, 365, 369, 380, 381, 421, 423, 452, 456, 459, 461, 463, 467, 481, 483], "weight_decai": [456, 459, 461, 467], "weight_fil": 5, "weights_fp16": 483, "well": [5, 324, 363, 375, 378, 483], "wen": 5, "went": 5, "were": [5, 487], "wet": 5, "what": [2, 5, 311], "whatsoev": 5, "whc": 337, "when": [0, 1, 2, 5, 7, 8, 94, 101, 112, 127, 183, 184, 186, 187, 188, 189, 191, 192, 194, 330, 331, 332, 333, 334, 335, 400, 404, 405, 421, 427, 433, 452, 454, 470, 476, 479, 480, 487], "where": [0, 6, 140, 172, 184, 237, 299, 303, 326, 327, 328, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 341, 343, 344, 348, 349, 350, 352, 355, 356, 362, 379, 382, 383, 397, 404, 405, 410, 411, 413, 424, 430, 436, 439, 441, 446, 463, 480, 481, 482], "wherea": 481, "whether": [143, 164, 186, 187, 192, 238, 343, 349, 362, 378, 383, 421, 424, 430], "which": [0, 1, 2, 5, 6, 7, 8, 18, 37, 83, 94, 101, 118, 121, 122, 125, 126, 127, 136, 145, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 165, 173, 174, 175, 176, 177, 179, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 194, 208, 219, 237, 241, 242, 256, 257, 259, 262, 263, 264, 265, 266, 278, 279, 287, 294, 299, 302, 303, 307, 327, 337, 338, 341, 356, 358, 362, 386, 421, 423, 426, 430, 433, 449, 462, 463, 476, 479, 480, 481, 482, 483, 487, 488], "while": [2, 3, 5, 8, 257, 386, 483, 484], "whistl": 2, "who": 5, "whose": [140, 307, 308], "why": 5, "wide": 483, "width": [327, 328, 331, 332, 334, 335, 337, 338, 356, 380, 381], "window": [8, 326, 327, 355, 356], "wipe": 8, "wire": 218, "wired_limit_mb": 218, "wise": [0, 2, 12, 13, 19, 20, 21, 22, 23, 24, 25, 87, 88, 89, 92, 106, 107, 128, 129, 133, 134, 135, 137, 139, 160, 161, 166, 167, 172, 180, 181, 182, 195, 196, 197, 198, 199, 200, 201, 202, 206, 222, 224, 226, 228, 234, 254, 255, 258, 261, 269, 270, 271, 272, 276, 277, 283, 289, 290, 329, 337, 338, 347, 357, 379, 390, 409, 416, 417, 419, 420, 435, 436, 438, 441, 442, 443, 444, 479], "wish": 8, "with_logit": 421, "within": [0, 3, 28, 172], "without": [1, 5, 7, 281, 378, 448, 478, 479, 480, 483, 484, 487], "wk": 5, "wl": 2, "wo": 5, "word": 0, "work": [2, 3, 5, 217, 479, 480, 481, 482, 483], "workhors": 324, "world": [314, 480], "worri": [1, 483], "would": [2, 5, 400, 480, 482, 483, 484, 487], "wq": 5, "wrap": [112, 324], "write": [0, 1, 2, 5, 324, 484], "written": 2, "wrt": 308, "wv": 5, "x": [0, 1, 2, 4, 5, 6, 38, 90, 112, 121, 122, 126, 127, 134, 139, 142, 143, 144, 164, 168, 169, 189, 238, 242, 247, 260, 265, 269, 297, 298, 304, 311, 313, 324, 326, 327, 328, 329, 339, 341, 342, 344, 348, 350, 351, 352, 355, 356, 357, 358, 379, 382, 384, 390, 391, 397, 400, 409, 410, 411, 412, 413, 414, 415, 416, 417, 418, 419, 420, 433, 435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445, 446, 447, 452, 454, 461, 479, 480, 481, 482, 483, 484, 485, 487], "x1": 422, "x2": 422, "x86_64": 8, "x_1": [422, 430], "x_2": [422, 430], "x_cast": 2, "x_grad": 1, "x_i": [420, 442, 443], "x_j": [442, 443], "x_offset": 2, "x_ptr": 2, "x_shape": 1, "x_stride": 2, "x_t": [343, 349, 383], "x_view": 484, "xcode": 8, "xcodeproj": 3, "xcrun": 8, "xf": 349, "xg": 349, "xi": 349, "xn": 343, "xo": 349, "xor": 89, "xr": 343, "xy": [0, 208], "xz": 343, "x\u00b2": 484, "y": [0, 2, 4, 5, 6, 38, 112, 168, 304, 324, 328, 337, 344, 348, 350, 352, 382, 425, 430, 433, 454, 457, 479, 480, 481, 483, 484], "y_": [425, 429], "y_cast": 2, "y_hat": 324, "y_offset": 2, "y_ptr": 2, "y_stride": 2, "ye": 5, "year": 5, "yet": [5, 189, 324, 452, 463, 481, 482, 483, 485], "yield": [5, 6, 476], "you": [2, 3, 5, 6, 7, 8, 218, 324, 391, 399, 449, 476, 479, 480, 481, 482, 484, 486, 487], "your": [2, 5, 8, 452, 481, 483], "z": [2, 343, 479, 483], "z_t": 343, "zeiler": 455, "zero": [0, 140, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 185, 208, 215, 296, 297, 298, 306, 324, 326, 327, 336, 337, 338, 365, 401, 402, 403, 404, 405, 406, 407, 408, 449, 454, 456, 482], "zero_grad": 481, "zeros_lik": 0, "zhang": 5, "zip": [5, 6], "zip_saf": 2}, "titles": ["Operations", "Custom Metal Kernels", "Custom Extensions in MLX", "Metal Debugger", "Linear Regression", "LLM inference", "Multi-Layer Perceptron", "MLX", "Build and Install", "mlx.core.Device", "mlx.core.Dtype", "mlx.core.DtypeCategory", "mlx.core.abs", "mlx.core.add", "mlx.core.addmm", "mlx.core.all", "mlx.core.allclose", "mlx.core.any", "mlx.core.arange", "mlx.core.arccos", "mlx.core.arccosh", "mlx.core.arcsin", "mlx.core.arcsinh", "mlx.core.arctan", "mlx.core.arctan2", "mlx.core.arctanh", "mlx.core.argmax", "mlx.core.argmin", "mlx.core.argpartition", "mlx.core.argsort", "mlx.core.array", "mlx.core.array.T", "mlx.core.array.abs", "mlx.core.array.all", "mlx.core.array.any", "mlx.core.array.argmax", "mlx.core.array.argmin", "mlx.core.array.astype", "mlx.core.array.at", "mlx.core.array.conj", "mlx.core.array.cos", "mlx.core.array.cummax", "mlx.core.array.cummin", "mlx.core.array.cumprod", "mlx.core.array.cumsum", "mlx.core.array.diag", "mlx.core.array.diagonal", "mlx.core.array.dtype", "mlx.core.array.exp", "mlx.core.array.flatten", "mlx.core.array.item", "mlx.core.array.itemsize", "mlx.core.array.log", "mlx.core.array.log10", "mlx.core.array.log1p", "mlx.core.array.log2", "mlx.core.array.logsumexp", "mlx.core.array.max", "mlx.core.array.mean", "mlx.core.array.min", "mlx.core.array.moveaxis", "mlx.core.array.nbytes", "mlx.core.array.ndim", "mlx.core.array.prod", "mlx.core.array.reciprocal", "mlx.core.array.reshape", "mlx.core.array.round", "mlx.core.array.rsqrt", "mlx.core.array.shape", "mlx.core.array.sin", "mlx.core.array.size", "mlx.core.array.split", "mlx.core.array.sqrt", "mlx.core.array.square", "mlx.core.array.squeeze", "mlx.core.array.std", "mlx.core.array.sum", "mlx.core.array.swapaxes", "mlx.core.array.tolist", "mlx.core.array.transpose", "mlx.core.array.var", "mlx.core.array.view", "mlx.core.array_equal", "mlx.core.as_strided", "mlx.core.atleast_1d", "mlx.core.atleast_2d", "mlx.core.atleast_3d", "mlx.core.bitwise_and", "mlx.core.bitwise_or", "mlx.core.bitwise_xor", "mlx.core.block_masked_mm", "mlx.core.broadcast_to", "mlx.core.ceil", "mlx.core.clip", "mlx.core.compile", "mlx.core.concatenate", "mlx.core.conj", "mlx.core.conjugate", "mlx.core.conv1d", "mlx.core.conv2d", "mlx.core.conv3d", "mlx.core.conv_general", "mlx.core.conv_transpose1d", "mlx.core.conv_transpose2d", "mlx.core.conv_transpose3d", "mlx.core.convolve", "mlx.core.cos", "mlx.core.cosh", "mlx.core.cummax", "mlx.core.cummin", "mlx.core.cumprod", "mlx.core.cumsum", "mlx.core.custom_function", "mlx.core.default_device", "mlx.core.default_stream", "mlx.core.degrees", "mlx.core.dequantize", "mlx.core.diag", "mlx.core.diagonal", "mlx.core.disable_compile", "mlx.core.distributed.Group", "mlx.core.distributed.all_gather", "mlx.core.distributed.all_sum", "mlx.core.distributed.init", "mlx.core.distributed.is_available", "mlx.core.distributed.recv", "mlx.core.distributed.recv_like", "mlx.core.distributed.send", "mlx.core.divide", "mlx.core.divmod", "mlx.core.einsum", "mlx.core.einsum_path", "mlx.core.enable_compile", "mlx.core.equal", "mlx.core.erf", "mlx.core.erfinv", "mlx.core.eval", "mlx.core.exp", "mlx.core.expand_dims", "mlx.core.expm1", "mlx.core.eye", "mlx.core.fast.affine_quantize", "mlx.core.fast.layer_norm", "mlx.core.fast.metal_kernel", "mlx.core.fast.rms_norm", "mlx.core.fast.rope", "mlx.core.fast.scaled_dot_product_attention", "mlx.core.fft.fft", "mlx.core.fft.fft2", "mlx.core.fft.fftn", "mlx.core.fft.ifft", "mlx.core.fft.ifft2", "mlx.core.fft.ifftn", "mlx.core.fft.irfft", "mlx.core.fft.irfft2", "mlx.core.fft.irfftn", "mlx.core.fft.rfft", "mlx.core.fft.rfft2", "mlx.core.fft.rfftn", "mlx.core.flatten", "mlx.core.floor", "mlx.core.floor_divide", "mlx.core.full", "mlx.core.gather_mm", "mlx.core.gather_qmm", "mlx.core.grad", "mlx.core.greater", "mlx.core.greater_equal", "mlx.core.hadamard_transform", "mlx.core.identity", "mlx.core.imag", "mlx.core.inner", "mlx.core.isclose", "mlx.core.isfinite", "mlx.core.isinf", "mlx.core.isnan", "mlx.core.isneginf", "mlx.core.isposinf", "mlx.core.issubdtype", "mlx.core.jvp", "mlx.core.left_shift", "mlx.core.less", "mlx.core.less_equal", "mlx.core.linalg.cholesky", "mlx.core.linalg.cholesky_inv", "mlx.core.linalg.cross", "mlx.core.linalg.eigh", "mlx.core.linalg.eigvalsh", "mlx.core.linalg.inv", "mlx.core.linalg.norm", "mlx.core.linalg.qr", "mlx.core.linalg.svd", "mlx.core.linalg.tri_inv", "mlx.core.linspace", "mlx.core.load", "mlx.core.log", "mlx.core.log10", "mlx.core.log1p", "mlx.core.log2", "mlx.core.logaddexp", "mlx.core.logical_and", "mlx.core.logical_not", "mlx.core.logical_or", "mlx.core.logsumexp", "mlx.core.matmul", "mlx.core.max", "mlx.core.maximum", "mlx.core.mean", "mlx.core.meshgrid", "mlx.core.metal.clear_cache", "mlx.core.metal.device_info", "mlx.core.metal.get_active_memory", "mlx.core.metal.get_cache_memory", "mlx.core.metal.get_peak_memory", "mlx.core.metal.is_available", "mlx.core.metal.reset_peak_memory", "mlx.core.metal.set_cache_limit", "mlx.core.metal.set_memory_limit", "mlx.core.metal.set_wired_limit", "mlx.core.metal.start_capture", "mlx.core.metal.stop_capture", "mlx.core.min", "mlx.core.minimum", "mlx.core.moveaxis", "mlx.core.multiply", "mlx.core.nan_to_num", "mlx.core.negative", "mlx.core.new_stream", "mlx.core.not_equal", "mlx.core.ones", "mlx.core.ones_like", "mlx.core.outer", "mlx.core.pad", "mlx.core.partition", "mlx.core.power", "mlx.core.prod", "mlx.core.put_along_axis", "mlx.core.quantize", "mlx.core.quantized_matmul", "mlx.core.radians", "mlx.core.random.bernoulli", "mlx.core.random.categorical", "mlx.core.random.gumbel", "mlx.core.random.key", "mlx.core.random.laplace", "mlx.core.random.multivariate_normal", "mlx.core.random.normal", "mlx.core.random.permutation", "mlx.core.random.randint", "mlx.core.random.seed", "mlx.core.random.split", "mlx.core.random.truncated_normal", "mlx.core.random.uniform", "mlx.core.real", "mlx.core.reciprocal", "mlx.core.remainder", "mlx.core.repeat", "mlx.core.reshape", "mlx.core.right_shift", "mlx.core.roll", "mlx.core.round", "mlx.core.rsqrt", "mlx.core.save", "mlx.core.save_gguf", "mlx.core.save_safetensors", "mlx.core.savez", "mlx.core.savez_compressed", "mlx.core.set_default_device", "mlx.core.set_default_stream", "mlx.core.sigmoid", "mlx.core.sign", "mlx.core.sin", "mlx.core.sinh", "mlx.core.softmax", "mlx.core.sort", "mlx.core.split", "mlx.core.sqrt", "mlx.core.square", "mlx.core.squeeze", "mlx.core.stack", "mlx.core.std", "mlx.core.stop_gradient", "mlx.core.stream", "mlx.core.subtract", "mlx.core.sum", "mlx.core.swapaxes", "mlx.core.synchronize", "mlx.core.take", "mlx.core.take_along_axis", "mlx.core.tan", "mlx.core.tanh", "mlx.core.tensordot", "mlx.core.tile", "mlx.core.topk", "mlx.core.trace", "mlx.core.transpose", "mlx.core.tri", "mlx.core.tril", "mlx.core.triu", "mlx.core.value_and_grad", "mlx.core.var", "mlx.core.view", "mlx.core.vjp", "mlx.core.vmap", "mlx.core.where", "mlx.core.zeros", "mlx.core.zeros_like", "mlx.nn.quantize", "mlx.nn.value_and_grad", "mlx.optimizers.clip_grad_norm", "mlx.utils.tree_flatten", "mlx.utils.tree_map", "mlx.utils.tree_map_with_path", "mlx.utils.tree_reduce", "mlx.utils.tree_unflatten", "mlx.core.Stream", "Array", "Data Types", "Devices and Streams", "Distributed Communication", "Fast", "FFT", "Linear Algebra", "Metal", "Neural Networks", "mlx.nn.ALiBi", "mlx.nn.AvgPool1d", "mlx.nn.AvgPool2d", "mlx.nn.BatchNorm", "mlx.nn.CELU", "mlx.nn.Conv1d", "mlx.nn.Conv2d", "mlx.nn.Conv3d", "mlx.nn.ConvTranspose1d", "mlx.nn.ConvTranspose2d", "mlx.nn.ConvTranspose3d", "mlx.nn.Dropout", "mlx.nn.Dropout2d", "mlx.nn.Dropout3d", "mlx.nn.ELU", "mlx.nn.Embedding", "mlx.nn.GELU", "mlx.nn.GLU", "mlx.nn.GRU", "mlx.nn.GroupNorm", "mlx.nn.HardShrink", "mlx.nn.HardTanh", "mlx.nn.Hardswish", "mlx.nn.InstanceNorm", "mlx.nn.LSTM", "mlx.nn.LayerNorm", "mlx.nn.LeakyReLU", "mlx.nn.Linear", "mlx.nn.LogSigmoid", "mlx.nn.LogSoftmax", "mlx.nn.MaxPool1d", "mlx.nn.MaxPool2d", "mlx.nn.Mish", "mlx.nn.Module.apply", "mlx.nn.Module.apply_to_modules", "mlx.nn.Module.children", "mlx.nn.Module.eval", "mlx.nn.Module.filter_and_map", "mlx.nn.Module.freeze", "mlx.nn.Module.leaf_modules", "mlx.nn.Module.load_weights", "mlx.nn.Module.modules", "mlx.nn.Module.named_modules", "mlx.nn.Module.parameters", "mlx.nn.Module.save_weights", "mlx.nn.Module.set_dtype", "mlx.nn.Module.state", "mlx.nn.Module.train", "mlx.nn.Module.trainable_parameters", "mlx.nn.Module.training", "mlx.nn.Module.unfreeze", "mlx.nn.Module.update", "mlx.nn.Module.update_modules", "mlx.nn.MultiHeadAttention", "mlx.nn.PReLU", "mlx.nn.QuantizedEmbedding", "mlx.nn.QuantizedLinear", "mlx.nn.RMSNorm", "mlx.nn.RNN", "mlx.nn.ReLU", "mlx.nn.ReLU6", "mlx.nn.RoPE", "mlx.nn.SELU", "mlx.nn.Sequential", "mlx.nn.SiLU", "mlx.nn.Sigmoid", "mlx.nn.SinusoidalPositionalEncoding", "mlx.nn.Softmax", "mlx.nn.Softmin", "mlx.nn.Softplus", "mlx.nn.Softshrink", "mlx.nn.Softsign", "mlx.nn.Step", "mlx.nn.Tanh", "mlx.nn.Transformer", "mlx.nn.Upsample", "mlx.nn.init.constant", "mlx.nn.init.glorot_normal", "mlx.nn.init.glorot_uniform", "mlx.nn.init.he_normal", "mlx.nn.init.he_uniform", "mlx.nn.init.identity", "mlx.nn.init.normal", "mlx.nn.init.uniform", "mlx.nn.celu", "mlx.nn.elu", "mlx.nn.gelu", "mlx.nn.gelu_approx", "mlx.nn.gelu_fast_approx", "mlx.nn.glu", "mlx.nn.hard_shrink", "mlx.nn.hard_tanh", "mlx.nn.hardswish", "mlx.nn.leaky_relu", "mlx.nn.log_sigmoid", "mlx.nn.log_softmax", "mlx.nn.losses.binary_cross_entropy", "mlx.nn.losses.cosine_similarity_loss", "mlx.nn.losses.cross_entropy", "mlx.nn.losses.gaussian_nll_loss", "mlx.nn.losses.hinge_loss", "mlx.nn.losses.huber_loss", "mlx.nn.losses.kl_div_loss", "mlx.nn.losses.l1_loss", "mlx.nn.losses.log_cosh_loss", "mlx.nn.losses.margin_ranking_loss", "mlx.nn.losses.mse_loss", "mlx.nn.losses.nll_loss", "mlx.nn.losses.smooth_l1_loss", "mlx.nn.losses.triplet_loss", "mlx.nn.mish", "mlx.nn.prelu", "mlx.nn.relu", "mlx.nn.relu6", "mlx.nn.selu", "mlx.nn.sigmoid", "mlx.nn.silu", "mlx.nn.softmax", "mlx.nn.softmin", "mlx.nn.softplus", "mlx.nn.softshrink", "mlx.nn.step", "mlx.nn.tanh", "Functions", "Initializers", "Layers", "Loss Functions", "Module", "Operations", "Optimizers", "mlx.optimizers.AdaDelta", "mlx.optimizers.Adafactor", "mlx.optimizers.Adagrad", "mlx.optimizers.Adam", "mlx.optimizers.AdamW", "mlx.optimizers.Adamax", "mlx.optimizers.Lion", "mlx.optimizers.Optimizer.apply_gradients", "mlx.optimizers.Optimizer.init", "mlx.optimizers.Optimizer.state", "mlx.optimizers.Optimizer.update", "mlx.optimizers.RMSprop", "mlx.optimizers.SGD", "mlx.optimizers.cosine_decay", "mlx.optimizers.exponential_decay", "mlx.optimizers.join_schedules", "mlx.optimizers.linear_schedule", "mlx.optimizers.step_decay", "Common Optimizers", "Optimizer", "Schedulers", "Random", "Transforms", "Tree Utils", "Compilation", "Distributed Communication", "Function Transforms", "Indexing Arrays", "Lazy Evaluation", "Conversion to NumPy and Other Frameworks", "Quick Start Guide", "Saving and Loading Arrays", "Unified Memory", "Using Streams"], "titleterms": {"A": 487, "In": 482, "The": 324, "ab": [12, 32], "adadelta": 455, "adafactor": 456, "adagrad": 457, "adam": 458, "adamax": 460, "adamw": 459, "add": 13, "addmm": 14, "affine_quant": 141, "algebra": 322, "alibi": 325, "all": [5, 15, 33, 480], "all_gath": 121, "all_sum": 122, "allclos": 16, "ani": [17, 34], "api": [7, 8], "appli": 358, "apply_gradi": 462, "apply_to_modul": 359, "arang": 18, "arcco": 19, "arccosh": 20, "arcsin": 21, "arcsinh": 22, "arctan": 23, "arctan2": 24, "arctanh": 25, "argmax": [26, 35], "argmin": [27, 36], "argpartit": 28, "argsort": 29, "arrai": [30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 316, 482, 486], "array_equ": 82, "as_strid": 83, "astyp": 37, "atleast_1d": 84, "atleast_2d": 85, "atleast_3d": 86, "attent": 5, "automat": 481, "avgpool1d": 326, "avgpool2d": 327, "back": 2, "basic": [479, 485], "batchnorm": 328, "benchmark": 5, "bernoulli": 240, "binari": 8, "binary_cross_entropi": 421, "bind": 2, "bitwise_and": 87, "bitwise_or": 88, "bitwise_xor": 89, "block_masked_mm": 90, "broadcast_to": 91, "build": [2, 8], "c": [7, 8], "categor": 241, "ceil": 92, "celu": [329, 409], "children": 360, "choleski": 183, "cholesky_inv": 184, "class": 324, "clear_cach": 209, "clip": 93, "clip_grad_norm": 309, "cmake": 2, "co": [40, 106], "code": [2, 5], "common": 473, "commun": [319, 480], "compil": [94, 479], "complex": 1, "comput": 483, "concaten": 95, "conj": [39, 96], "conjug": 97, "constant": 401, "conv1d": [98, 330], "conv2d": [99, 331], "conv3d": [100, 332], "conv_gener": 101, "conv_transpose1d": 102, "conv_transpose2d": 103, "conv_transpose3d": 104, "convers": 484, "convert": 5, "convolv": 105, "convtranspose1d": 333, "convtranspose2d": 334, "convtranspose3d": 335, "core": [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 267, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 315], "cosh": 107, "cosine_decai": 468, "cosine_similarity_loss": 422, "cpu": 2, "cross": 185, "cross_entropi": 423, "cummax": [41, 108], "cummin": [42, 109], "cumprod": [43, 110], "cumsum": [44, 111], "custom": [1, 2], "custom_funct": 112, "data": 317, "debug": 479, "debugg": 3, "default_devic": 113, "default_stream": 114, "degre": 115, "dequant": 116, "devic": [9, 318], "device_info": 210, "diag": [45, 117], "diagon": [46, 118], "differ": 482, "differenti": 481, "disable_compil": 119, "distribut": [120, 121, 122, 123, 124, 125, 126, 127, 319, 480], "divid": 128, "divmod": 129, "download": [2, 5], "dropout": 336, "dropout2d": 337, "dropout3d": 338, "dtype": [10, 47], "dtypecategori": 11, "eigh": 186, "eigvalsh": 187, "einsum": 130, "einsum_path": 131, "elu": [339, 410], "embed": 340, "enable_compil": 132, "encod": 5, "end": 2, "equal": 133, "erf": 134, "erfinv": 135, "eval": [136, 361], "evalu": 483, "exampl": [1, 2, 7, 479, 480, 487], "exp": [48, 137], "expand_dim": 138, "expm1": 139, "exponential_decai": 469, "extens": 2, "ey": 140, "fast": [141, 142, 143, 144, 145, 146, 320], "fft": [147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 321], "fft2": 148, "fftn": 149, "filter_and_map": 362, "flatten": [49, 159], "floor": 160, "floor_divid": 161, "format": 486, "found": 8, "framework": 484, "freez": 363, "from": [8, 482], "full": [5, 162], "function": [448, 451, 479, 481, 485], "further": 7, "gather_mm": 163, "gather_qmm": 164, "gaussian_nll_loss": 424, "gelu": [341, 411], "gelu_approx": 412, "gelu_fast_approx": 413, "gener": 5, "get": 480, "get_active_memori": 211, "get_cache_memori": 212, "get_peak_memori": 213, "glorot_norm": 402, "glorot_uniform": 403, "glu": [342, 414], "gpu": 2, "grad": [165, 324], "graph": [479, 483, 485], "greater": 166, "greater_equ": 167, "grid": 1, "group": 120, "groupnorm": 344, "gru": 343, "guid": 485, "gumbel": 242, "hadamard_transform": 168, "hard_shrink": 415, "hard_tanh": 416, "hardshrink": 345, "hardswish": [347, 417], "hardtanh": 346, "he_norm": 404, "he_uniform": 405, "hinge_loss": 425, "host": 480, "huber_loss": 426, "ident": [169, 406], "ifft": 150, "ifft2": 151, "ifftn": 152, "imag": 170, "implement": [2, 5], "index": 482, "infer": 5, "init": [123, 401, 402, 403, 404, 405, 406, 407, 408, 463], "initi": 449, "inner": 171, "inspect": 324, "instal": [7, 8, 480], "instancenorm": 348, "introduc": 2, "inv": 188, "irfft": 153, "irfft2": 154, "irfftn": 155, "is_avail": [124, 214], "isclos": 172, "isfinit": 173, "isinf": 174, "isnan": 175, "isneginf": 176, "isposinf": 177, "issubdtyp": 178, "item": 50, "items": 51, "jax": 484, "join_schedul": 470, "jvp": 179, "kei": 243, "kernel": 1, "kl_div_loss": 427, "l1_loss": 428, "laplac": 244, "layer": [5, 6, 450], "layer_norm": 142, "layernorm": 350, "lazi": 483, "leaf_modul": 364, "leaky_relu": 418, "leakyrelu": 351, "left_shift": 180, "less": 181, "less_equ": 182, "linalg": [183, 184, 185, 186, 187, 188, 189, 190, 191, 192], "linear": [4, 322, 352], "linear_schedul": 471, "linspac": 193, "lion": 461, "llm": 5, "load": [5, 194, 454, 486], "load_weight": 365, "log": [52, 195], "log10": [53, 196], "log1p": [54, 197], "log2": [55, 198], "log_cosh_loss": 429, "log_sigmoid": 419, "log_softmax": 420, "logaddexp": 199, "logical_and": 200, "logical_not": 201, "logical_or": 202, "logsigmoid": 353, "logsoftmax": 354, "logsumexp": [56, 203], "loss": [421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 451], "lstm": 349, "margin_ranking_loss": 430, "matmul": 204, "max": [57, 205], "maximum": 206, "maxpool1d": 355, "maxpool2d": 356, "mean": [58, 207], "memori": 487, "meshgrid": 208, "metal": [1, 3, 8, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 323], "metal_kernel": 143, "min": [59, 221], "minim": 8, "minimum": 222, "mish": [357, 435], "mlx": [2, 7, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 267, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 288, 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 307, 308, 309, 310, 311, 312, 313, 314, 315, 325, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350, 351, 352, 353, 354, 355, 356, 357, 358, 359, 360, 361, 362, 363, 364, 365, 366, 367, 368, 369, 370, 371, 372, 373, 374, 375, 376, 377, 378, 379, 380, 381, 382, 383, 384, 385, 386, 387, 388, 389, 390, 391, 392, 393, 394, 395, 396, 397, 398, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 409, 410, 411, 412, 413, 414, 415, 416, 417, 418, 419, 420, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445, 446, 447, 455, 456, 457, 458, 459, 460, 461, 462, 463, 464, 465, 466, 467, 468, 469, 470, 471, 472], "model": 5, "modul": [324, 358, 359, 360, 361, 362, 363, 364, 365, 366, 367, 368, 369, 370, 371, 372, 373, 374, 375, 376, 377, 452], "moveaxi": [60, 223], "mpi": 480, "mse_loss": 431, "multi": 6, "multiheadattent": 378, "multipli": 224, "multivariate_norm": 245, "named_modul": 367, "nan_to_num": 225, "nbyte": 61, "ndim": 62, "neg": 226, "network": 324, "neural": 324, "new_stream": 227, "nll_loss": 432, "nn": [307, 308, 325, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350, 351, 352, 353, 354, 355, 356, 357, 358, 359, 360, 361, 362, 363, 364, 365, 366, 367, 368, 369, 370, 371, 372, 373, 374, 375, 376, 377, 378, 379, 380, 381, 382, 383, 384, 385, 386, 387, 388, 389, 390, 391, 392, 393, 394, 395, 396, 397, 398, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 409, 410, 411, 412, 413, 414, 415, 416, 417, 418, 419, 420, 421, 422, 423, 424, 425, 426, 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445, 446, 447], "norm": 189, "normal": [246, 407], "not_equ": 228, "numpi": [482, 484], "ones": 229, "ones_lik": 230, "onli": 483, "oper": [0, 2, 453], "optim": [309, 454, 455, 456, 457, 458, 459, 460, 461, 462, 463, 464, 465, 466, 467, 468, 469, 470, 471, 472, 473, 474], "option": 8, "other": 484, "outer": 231, "pad": 232, "paramet": [324, 368], "partit": 233, "perceptron": 6, "permut": 247, "place": 482, "power": 234, "prelu": [379, 436], "primit": 2, "prod": [63, 235], "pure": 479, "put": 5, "put_along_axi": 236, "python": [2, 7, 8], "pytorch": 484, "qr": 190, "quantiz": [237, 307], "quantized_matmul": 238, "quantizedembed": 380, "quantizedlinear": 381, "quick": [324, 485], "radian": 239, "randint": 248, "random": [240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 476], "read": 7, "real": 253, "reciproc": [64, 254], "recv": 125, "recv_lik": 126, "reduc": 480, "refer": 7, "regress": 4, "relu": [384, 437], "relu6": [385, 438], "remaind": 255, "remot": 480, "repeat": 256, "requir": 8, "reset_peak_memori": 215, "reshap": [65, 257], "result": 2, "rfft": 156, "rfft2": 157, "rfftn": 158, "right_shift": 258, "rms_norm": 144, "rmsnorm": 382, "rmsprop": 466, "rnn": 383, "roll": 259, "rope": [145, 386], "round": [66, 260], "rsqrt": [67, 261], "sampl": 1, "save": [262, 454, 486], "save_gguf": 263, "save_safetensor": 264, "save_weight": 369, "savez": 265, "savez_compress": 266, "scaled_dot_product_attent": 146, "schedul": 475, "script": [2, 5], "seed": 249, "selu": [387, 439], "send": 127, "sequenti": 388, "serial": 486, "set": 480, "set_cache_limit": 216, "set_default_devic": 267, "set_default_stream": 268, "set_dtyp": 370, "set_memory_limit": 217, "set_wired_limit": 218, "setuptool": 2, "sgd": 467, "shape": [1, 68], "shell": 8, "sigmoid": [269, 390, 440], "sign": 270, "silu": [389, 441], "simpl": [1, 487], "sin": [69, 271], "sinh": 272, "sinusoidalpositionalencod": 391, "size": [8, 70], "smooth_l1_loss": 433, "softmax": [273, 392, 442], "softmin": [393, 443], "softplu": [394, 444], "softshrink": [395, 445], "softsign": 396, "sort": 274, "sourc": 8, "specifi": 488, "speedup": 479, "split": [71, 250, 275], "sqrt": [72, 276], "squar": [73, 277], "squeez": [74, 278], "stack": 279, "start": [324, 480, 485], "start_captur": 219, "state": [371, 464], "std": [75, 280], "step": [397, 446], "step_decai": 472, "stop_captur": 220, "stop_gradi": 281, "stream": [282, 315, 318, 488], "stride": 1, "subtract": 283, "sum": [76, 284], "support": 317, "svd": 191, "swapax": [77, 285], "synchron": 286, "t": 31, "take": 287, "take_along_axi": 288, "tan": 289, "tanh": [290, 398, 447], "tensordot": 291, "tensorflow": 484, "tile": 292, "togeth": 5, "tolist": 78, "topk": 293, "trace": 294, "train": [372, 374, 479, 480], "trainable_paramet": 373, "transform": [2, 399, 477, 479, 481, 483, 485], "transpos": [79, 295], "tree": 478, "tree_flatten": 310, "tree_map": 311, "tree_map_with_path": 312, "tree_reduc": 313, "tree_unflatten": 314, "tri": 296, "tri_inv": 192, "tril": 297, "triplet_loss": 434, "triu": 298, "troubleshoot": 8, "truncated_norm": 251, "tune": 480, "type": 317, "unfreez": 375, "unifi": 487, "uniform": [252, 408], "up": 480, "updat": [324, 376, 465, 482], "update_modul": 377, "upsampl": 400, "us": [1, 2, 483, 488], "usag": [2, 7], "util": [310, 311, 312, 313, 314, 478], "valu": 324, "value_and_grad": [299, 308], "var": [80, 300], "vector": 481, "view": [81, 301], "vjp": [1, 302], "vmap": 303, "weight": 5, "what": 483, "when": 483, "where": 304, "why": 483, "workflow": 3, "x86": 8, "xcode": 3, "you": 483, "zero": 305, "zeros_lik": 306}}) \ No newline at end of file diff --git a/docs/build/html/structmlx_1_1core_1_1metal_1_1_command_encoder-members.html b/docs/build/html/structmlx_1_1core_1_1metal_1_1_command_encoder-members.html index 53b080582..981601231 100644 --- a/docs/build/html/structmlx_1_1core_1_1metal_1_1_command_encoder-members.html +++ b/docs/build/html/structmlx_1_1core_1_1metal_1_1_command_encoder-members.html @@ -99,13 +99,14 @@ $(function(){ initResizable(false); }); dispatchThreadgroups(MTL::Size grid_dims, MTL::Size group_dims)mlx::core::metal::CommandEncoder dispatchThreads(MTL::Size grid_dims, MTL::Size group_dims)mlx::core::metal::CommandEncoder inputs()mlx::core::metal::CommandEncoderinline - operator->()mlx::core::metal::CommandEncoderinline - operator=(const CommandEncoder &)=deletemlx::core::metal::CommandEncoder - outputs()mlx::core::metal::CommandEncoderinline - set_input_array(const array &a, int idx, int64_t offset=0)mlx::core::metal::CommandEncoder - set_output_array(array &a, int idx, int64_t offset=0)mlx::core::metal::CommandEncoder - start_concurrent()mlx::core::metal::CommandEncoderinline - ~CommandEncoder()mlx::core::metal::CommandEncoder + maybeInsertBarrier()mlx::core::metal::CommandEncoder + operator->()mlx::core::metal::CommandEncoderinline + operator=(const CommandEncoder &)=deletemlx::core::metal::CommandEncoder + outputs()mlx::core::metal::CommandEncoderinline + set_input_array(const array &a, int idx, int64_t offset=0)mlx::core::metal::CommandEncoder + set_output_array(array &a, int idx, int64_t offset=0)mlx::core::metal::CommandEncoder + start_concurrent()mlx::core::metal::CommandEncoderinline + ~CommandEncoder()mlx::core::metal::CommandEncoder
    diff --git a/docs/build/html/structmlx_1_1core_1_1metal_1_1_command_encoder.html b/docs/build/html/structmlx_1_1core_1_1metal_1_1_command_encoder.html index 186308cbb..ee359fde9 100644 --- a/docs/build/html/structmlx_1_1core_1_1metal_1_1_command_encoder.html +++ b/docs/build/html/structmlx_1_1core_1_1metal_1_1_command_encoder.html @@ -121,6 +121,8 @@ Public Member Functions   void dispatchThreads (MTL::Size grid_dims, MTL::Size group_dims)   +void maybeInsertBarrier () +  ConcurrentContext start_concurrent ()    ~CommandEncoder () @@ -256,6 +258,23 @@ Public Member Functions
    +
    +
    + +

    ◆ maybeInsertBarrier()

    + +
    +
    + + + + + + + +
    void mlx::core::metal::CommandEncoder::maybeInsertBarrier ()
    +
    +
    diff --git a/docs/build/html/usage/function_transforms.html b/docs/build/html/usage/function_transforms.html index 1b08af1eb..e903d77a8 100644 --- a/docs/build/html/usage/function_transforms.html +++ b/docs/build/html/usage/function_transforms.html @@ -986,13 +986,13 @@ We will prioritize including it.

    ys = mx.random.uniform(shape=(100, 4096)) def naive_add(xs, ys): - return [xs[i] + ys[:, i] for i in range(xs.shape[1])] + return [xs[i] + ys[:, i] for i in range(xs.shape[0])]

    Instead you can use vmap() to automatically vectorize the addition:

    # Vectorize over the second dimension of x and the
 # first dimension of y
-vmap_add = mx.vmap(lambda x, y: x + y, in_axes=(1, 0))
+vmap_add = mx.vmap(lambda x, y: x + y, in_axes=(0, 1))

    The in_axes parameter can be used to specify which dimensions of the diff --git a/docs/build/html/usage/indexing.html b/docs/build/html/usage/indexing.html index 37d46eff8..55e3de841 100644 --- a/docs/build/html/usage/indexing.html +++ b/docs/build/html/usage/indexing.html @@ -922,7 +922,7 @@ undefined behavior.

    from the GPU. Performing bounds checking for array indices before launching the kernel would be extremely inefficient.

    Indexing with boolean masks is something that MLX may support in the future. In -general, MLX has limited support for operations for which outputs +general, MLX has limited support for operations for which output shapes are dependent on input data. Other examples of these types of operations which MLX does not yet support include numpy.nonzero() and the single input version of numpy.where().

    diff --git a/docs/build/html/usage/lazy_evaluation.html b/docs/build/html/usage/lazy_evaluation.html index cac88c1c1..59143b41a 100644 --- a/docs/build/html/usage/lazy_evaluation.html +++ b/docs/build/html/usage/lazy_evaluation.html @@ -952,7 +952,7 @@ stochastic gradient descent). A natural and usually efficient place to use

    An important behavior to be aware of is when the graph will be implicitly evaluated. Anytime you print an array, convert it to an -numpy.ndarray, or otherwise access it’s memory via memoryview, +numpy.ndarray, or otherwise access its memory via memoryview, the graph will be evaluated. Saving arrays via save() (or any other MLX saving functions) will also evaluate the array.

    Calling array.item() on a scalar array will also evaluate it. In the