mlx/python/src/array.cpp

// Copyright © 2023 Apple Inc.

#include <cstdint>
#include <cstring>
#include <sstream>

#include <pybind11/numpy.h>

#include "python/src/indexing.h"
#include "python/src/utils.h"

#include "mlx/ops.h"
#include "mlx/transforms.h"
#include "mlx/utils.h"

namespace py = pybind11;
using namespace py::literals;

enum PyScalarT {
  pybool = 0,
  pyint = 1,
  pyfloat = 2,
  pycomplex = 3,
};

template <typename T>
py::list to_list(array& a, size_t index, int dim) {
  py::list pl;
  auto stride = a.strides()[dim];
  for (int i = 0; i < a.shape(dim); ++i) {
    if (dim == a.ndim() - 1) {
      pl.append((a.data<T>()[index]));
    } else {
      pl.append(to_list<T>(a, index, dim + 1));
    }
    index += stride;
  }
  return pl;
}

auto to_scalar(array& a) {
  {
    py::gil_scoped_release nogil;
    a.eval();
  }
  switch (a.dtype()) {
    case bool_:
      return py::cast(a.item<bool>());
    case uint8:
      return py::cast(a.item<uint8_t>());
    case uint16:
      return py::cast(a.item<uint16_t>());
    case uint32:
      return py::cast(a.item<uint32_t>());
    case uint64:
      return py::cast(a.item<uint64_t>());
    case int8:
      return py::cast(a.item<int8_t>());
    case int16:
      return py::cast(a.item<int16_t>());
    case int32:
      return py::cast(a.item<int32_t>());
    case int64:
      return py::cast(a.item<int64_t>());
    case float16:
      return py::cast(static_cast<float>(a.item<float16_t>()));
    case float32:
      return py::cast(a.item<float>());
    case bfloat16:
      return py::cast(static_cast<float>(a.item<bfloat16_t>()));
    case complex64:
      return py::cast(a.item<std::complex<float>>());
  }
}

py::object tolist(array& a) {
  if (a.ndim() == 0) {
    return to_scalar(a);
  }
  {
    py::gil_scoped_release nogil;
    a.eval();
  }
  py::object pl;
  switch (a.dtype()) {
    case bool_:
      return to_list<bool>(a, 0, 0);
    case uint8:
      return to_list<uint8_t>(a, 0, 0);
    case uint16:
      return to_list<uint16_t>(a, 0, 0);
    case uint32:
      return to_list<uint32_t>(a, 0, 0);
    case uint64:
      return to_list<uint64_t>(a, 0, 0);
    case int8:
      return to_list<int8_t>(a, 0, 0);
    case int16:
      return to_list<int16_t>(a, 0, 0);
    case int32:
      return to_list<int32_t>(a, 0, 0);
    case int64:
      return to_list<int64_t>(a, 0, 0);
    case float16:
      return to_list<float16_t>(a, 0, 0);
    case float32:
      return to_list<float>(a, 0, 0);
    case bfloat16:
      return to_list<float16_t>(a, 0, 0);
    case complex64:
      return to_list<std::complex<float>>(a, 0, 0);
  }
}

template <typename T, typename U>
void fill_vector(T list, std::vector<U>& vals) {
  for (auto l : list) {
    if (py::isinstance<py::list>(l)) {
      fill_vector(l.template cast<py::list>(), vals);
    } else if (py::isinstance<py::tuple>(*list.begin())) {
      fill_vector(l.template cast<py::tuple>(), vals);
    } else {
      vals.push_back(l.template cast<U>());
    }
  }
}

template <typename T>
PyScalarT validate_shape(
    T list,
    const std::vector<int>& shape,
    int idx,
    bool& all_python_primitive_elements) {
  if (idx >= shape.size()) {
    throw std::invalid_argument("Initialization encountered extra dimension.");
  }
  auto s = shape[idx];
  if (py::len(list) != s) {
    throw std::invalid_argument(
        "Initialization encountered non-uniform length.");
  }

  if (s == 0) {
    return pyfloat;
  }

  PyScalarT type = pybool;
  for (auto l : list) {
    PyScalarT t;
    if (py::isinstance<py::list>(l)) {
      t = validate_shape(
          l.template cast<py::list>(),
          shape,
          idx + 1,
          all_python_primitive_elements);
    } else if (py::isinstance<py::tuple>(*list.begin())) {
      t = validate_shape(
          l.template cast<py::tuple>(),
          shape,
          idx + 1,
          all_python_primitive_elements);
    } else if (py::isinstance<py::bool_>(l)) {
      t = pybool;
    } else if (py::isinstance<py::int_>(l)) {
      t = pyint;
    } else if (py::isinstance<py::float_>(l)) {
      t = pyfloat;
    } else if (PyComplex_Check(l.ptr())) {
      t = pycomplex;
    } else if (py::isinstance<array>(l)) {
      all_python_primitive_elements = false;
      auto arr = py::cast<array>(l);
      if (arr.ndim() + idx + 1 == shape.size() &&
          std::equal(
              arr.shape().cbegin(),
              arr.shape().cend(),
              shape.cbegin() + idx + 1)) {
        t = pybool;
      } else {
        throw std::invalid_argument(
            "Initialization encountered non-uniform length.");
      }
    } else {
      std::ostringstream msg;
      msg << "Invalid type in array initialization" << l.get_type() << ".";
      throw std::invalid_argument(msg.str());
    }
    type = std::max(type, t);
  }
  return type;
}

template <typename T>
void get_shape(T list, std::vector<int>& shape) {
  shape.push_back(py::len(list));
  if (shape.back() > 0) {
    auto& l = *list.begin();
    if (py::isinstance<py::list>(l)) {
      return get_shape(l.template cast<py::list>(), shape);
    } else if (py::isinstance<py::tuple>(l)) {
      return get_shape(l.template cast<py::tuple>(), shape);
    } else if (py::isinstance<array>(l)) {
      auto arr = py::cast<array>(l);
      shape.insert(shape.end(), arr.shape().begin(), arr.shape().end());
      return;
    }
  }
}

using array_init_type = std::variant<
    py::bool_,
    py::int_,
    py::float_,
    std::complex<float>,
    py::list,
    py::tuple,
    array,
    py::array,
    py::buffer,
    py::object>;

// Forward declaration
array create_array(array_init_type v, std::optional<Dtype> t);

template <typename T>
array array_from_list(
    T pl,
    const PyScalarT& inferred_type,
    std::optional<Dtype> specified_type,
    const std::vector<int>& shape) {
  // Make the array
  switch (inferred_type) {
    case pybool: {
      std::vector<bool> vals;
      fill_vector(pl, vals);
      return array(vals.begin(), shape, specified_type.value_or(bool_));
    }
    case pyint: {
      auto dtype = specified_type.value_or(int32);
      if (dtype == int64) {
        std::vector<int64_t> vals;
        fill_vector(pl, vals);
        return array(vals.begin(), shape, dtype);
      } else if (dtype == uint64) {
        std::vector<uint64_t> vals;
        fill_vector(pl, vals);
        return array(vals.begin(), shape, dtype);
      } else if (dtype == uint32) {
        std::vector<uint32_t> vals;
        fill_vector(pl, vals);
        return array(vals.begin(), shape, dtype);
      } else if (is_floating_point(dtype)) {
        std::vector<float> vals;
        fill_vector(pl, vals);
        return array(vals.begin(), shape, dtype);
      } else {
        std::vector<int> vals;
        fill_vector(pl, vals);
        return array(vals.begin(), shape, dtype);
      }
    }
    case pyfloat: {
      std::vector<float> vals;
      fill_vector(pl, vals);
      return array(vals.begin(), shape, specified_type.value_or(float32));
    }
    case pycomplex: {
      std::vector<std::complex<float>> vals;
      fill_vector(pl, vals);
      return array(
          reinterpret_cast<complex64_t*>(vals.data()),
          shape,
          specified_type.value_or(complex64));
    }
    default: {
      std::ostringstream msg;
      msg << "Should not happen, inferred: " << inferred_type
          << " on subarray made of only python primitive types.";
      throw std::runtime_error(msg.str());
    }
  }
}

template <typename T>
array array_from_list(T pl, std::optional<Dtype> dtype) {
  // Compute the shape
  std::vector<int> shape;
  get_shape(pl, shape);

  // Validate the shape and type
  bool all_python_primitive_elements = true;
  auto type = validate_shape(pl, shape, 0, all_python_primitive_elements);

  if (all_python_primitive_elements) {
    // `pl` does not contain mlx arrays
    return array_from_list(pl, type, dtype, shape);
  }

  // `pl` contains mlx arrays
  std::vector<array> arrays;
  for (auto l : pl) {
    arrays.push_back(create_array(py::cast<array_init_type>(l), dtype));
  }
  return stack(arrays);
}

///////////////////////////////////////////////////////////////////////////////
// Numpy -> MLX
///////////////////////////////////////////////////////////////////////////////

template <typename T>
array np_array_to_mlx_contiguous(
    py::array_t<T, py::array::c_style | py::array::forcecast> np_array,
    const std::vector<int>& shape,
    Dtype dtype) {
  // Make a copy of the numpy buffer
  // Get buffer ptr pass to array constructor
  py::buffer_info buf = np_array.request();
  const T* data_ptr = static_cast<T*>(buf.ptr);
  return array(data_ptr, shape, dtype);

  // Note: Leaving the following memoryless copy from np to mx commented
  // out for the time being since it is unsafe given that the incoming
  // numpy array may change the underlying data

  // // Share underlying numpy buffer
  // // Copy to increase ref count
  // auto deleter = [np_array](void*) {};
  // void* data_ptr = np_array.mutable_data();
  // // Use buffer from numpy
  // return array(data_ptr, deleter, shape, dtype);
}

template <>
array np_array_to_mlx_contiguous(
    py::array_t<std::complex<float>, py::array::c_style | py::array::forcecast>
        np_array,
    const std::vector<int>& shape,
    Dtype dtype) {
  // Get buffer ptr pass to array constructor
  py::buffer_info buf = np_array.request();
  auto data_ptr = static_cast<std::complex<float>*>(buf.ptr);
  return array(reinterpret_cast<complex64_t*>(data_ptr), shape, dtype);
}

array np_array_to_mlx(py::array np_array, std::optional<Dtype> dtype) {
  // Compute the shape and size
  std::vector<int> shape;
  for (int i = 0; i < np_array.ndim(); i++) {
    shape.push_back(np_array.shape(i));
  }

  // Get dtype
  auto type = np_array.dtype();

  // Copy data and make array
  if (type.is(py::dtype::of<int>())) {
    return np_array_to_mlx_contiguous<int32_t>(
        np_array, shape, dtype.value_or(int32));
  } else if (type.is(py::dtype::of<uint32_t>())) {
    return np_array_to_mlx_contiguous<uint32_t>(
        np_array, shape, dtype.value_or(uint32));
  } else if (type.is(py::dtype::of<bool>())) {
    return np_array_to_mlx_contiguous<bool>(
        np_array, shape, dtype.value_or(bool_));
  } else if (type.is(py::dtype::of<double>())) {
    return np_array_to_mlx_contiguous<double>(
        np_array, shape, dtype.value_or(float32));
  } else if (type.is(py::dtype::of<float>())) {
    return np_array_to_mlx_contiguous<float>(
        np_array, shape, dtype.value_or(float32));
  } else if (type.is(py::dtype("float16"))) {
    return np_array_to_mlx_contiguous<float>(
        np_array, shape, dtype.value_or(float16));
  } else if (type.is(py::dtype::of<uint8_t>())) {
    return np_array_to_mlx_contiguous<uint8_t>(
        np_array, shape, dtype.value_or(uint8));
  } else if (type.is(py::dtype::of<uint16_t>())) {
    return np_array_to_mlx_contiguous<uint16_t>(
        np_array, shape, dtype.value_or(uint16));
  } else if (type.is(py::dtype::of<uint64_t>())) {
    return np_array_to_mlx_contiguous<uint64_t>(
        np_array, shape, dtype.value_or(uint64));
  } else if (type.is(py::dtype::of<int8_t>())) {
    return np_array_to_mlx_contiguous<int8_t>(
        np_array, shape, dtype.value_or(int8));
  } else if (type.is(py::dtype::of<int16_t>())) {
    return np_array_to_mlx_contiguous<int16_t>(
        np_array, shape, dtype.value_or(int16));
  } else if (type.is(py::dtype::of<int64_t>())) {
    return np_array_to_mlx_contiguous<int64_t>(
        np_array, shape, dtype.value_or(int64));
  } else if (type.is(py::dtype::of<std::complex<float>>())) {
    return np_array_to_mlx_contiguous<std::complex<float>>(
        np_array, shape, dtype.value_or(complex64));
  } else if (type.is(py::dtype::of<std::complex<double>>())) {
    return np_array_to_mlx_contiguous<std::complex<float>>(
        np_array, shape, dtype.value_or(complex64));
  } else {
    std::ostringstream msg;
    msg << "Cannot convert numpy array of type " << type << " to mlx array.";
    throw std::invalid_argument(msg.str());
  }
}

///////////////////////////////////////////////////////////////////////////////
// Python Buffer Protocol (MLX -> Numpy)
///////////////////////////////////////////////////////////////////////////////

std::optional<std::string> buffer_format(const array& a) {
  // https://docs.python.org/3.10/library/struct.html#format-characters
  switch (a.dtype()) {
    case bool_:
      return pybind11::format_descriptor<bool>::format();
    case uint8:
      return pybind11::format_descriptor<uint8_t>::format();
    case uint16:
      return pybind11::format_descriptor<uint16_t>::format();
    case uint32:
      return pybind11::format_descriptor<uint32_t>::format();
    case uint64:
      return pybind11::format_descriptor<uint64_t>::format();
    case int8:
      return pybind11::format_descriptor<int8_t>::format();
    case int16:
      return pybind11::format_descriptor<int16_t>::format();
    case int32:
      return pybind11::format_descriptor<int32_t>::format();
    case int64:
      return pybind11::format_descriptor<int64_t>::format();
    case float16:
      // https://github.com/pybind/pybind11/issues/4998
      return "e";
    case float32: {
      return pybind11::format_descriptor<float>::format();
    }
    case bfloat16:
      // not supported by python buffer protocol or numpy.
      // must be null according to
      // https://docs.python.org/3.10/c-api/buffer.html#c.PyBUF_FORMAT
      // which implies 'B'.
      return {};
    case complex64:
      return pybind11::format_descriptor<std::complex<float>>::format();
    default: {
      std::ostringstream os;
      os << "bad dtype: " << a.dtype();
      throw std::runtime_error(os.str());
    }
  }
}

std::vector<size_t> buffer_strides(const array& a) {
  std::vector<size_t> py_strides;
  py_strides.reserve(a.strides().size());
  for (const size_t stride : a.strides()) {
    py_strides.push_back(stride * a.itemsize());
  }
  return py_strides;
}

///////////////////////////////////////////////////////////////////////////////
// Module
///////////////////////////////////////////////////////////////////////////////

array create_array(array_init_type v, std::optional<Dtype> t) {
  if (auto pv = std::get_if<py::bool_>(&v); pv) {
    return array(py::cast<bool>(*pv), t.value_or(bool_));
  } else if (auto pv = std::get_if<py::int_>(&v); pv) {
    return array(py::cast<int>(*pv), t.value_or(int32));
  } else if (auto pv = std::get_if<py::float_>(&v); pv) {
    return array(py::cast<float>(*pv), t.value_or(float32));
  } else if (auto pv = std::get_if<std::complex<float>>(&v); pv) {
    return array(static_cast<complex64_t>(*pv), t.value_or(complex64));
  } else if (auto pv = std::get_if<py::list>(&v); pv) {
    return array_from_list(*pv, t);
  } else if (auto pv = std::get_if<py::tuple>(&v); pv) {
    return array_from_list(*pv, t);
  } else if (auto pv = std::get_if<array>(&v); pv) {
    return astype(*pv, t.value_or((*pv).dtype()));
  } else if (auto pv = std::get_if<py::array>(&v); pv) {
    return np_array_to_mlx(*pv, t);
  } else if (auto pv = std::get_if<py::buffer>(&v); pv) {
    return np_array_to_mlx(*pv, t);
  } else {
    auto arr = to_array_with_accessor(std::get<py::object>(v));
    return astype(arr, t.value_or(arr.dtype()));
  }
}

class ArrayAt {
 public:
  ArrayAt(array x) : x_(std::move(x)) {}
  ArrayAt& set_indices(py::object indices) {
    indices_ = indices;
    return *this;
  }
  array add(const ScalarOrArray& v) {
    return mlx_add_item(x_, indices_, v);
  }
  array subtract(const ScalarOrArray& v) {
    return mlx_subtract_item(x_, indices_, v);
  }
  array multiply(const ScalarOrArray& v) {
    return mlx_multiply_item(x_, indices_, v);
  }
  array divide(const ScalarOrArray& v) {
    return mlx_divide_item(x_, indices_, v);
  }
  array maximum(const ScalarOrArray& v) {
    return mlx_maximum_item(x_, indices_, v);
  }
  array minimum(const ScalarOrArray& v) {
    return mlx_minimum_item(x_, indices_, v);
  }

 private:
  array x_;
  py::object indices_;
};

class ArrayPythonIterator {
 public:
  ArrayPythonIterator(array x) : idx_(0), x_(std::move(x)) {
    if (x_.shape(0) > 0 && x_.shape(0) < 10) {
      splits_ = split(x_, x_.shape(0));
    }
  }

  array next() {
    if (idx_ >= x_.shape(0)) {
      throw py::stop_iteration();
    }

    if (idx_ >= 0 && idx_ < splits_.size()) {
      return squeeze(splits_[idx_++], 0);
    }

    return *(x_.begin() + idx_++);
  }

 private:
  int idx_;
  array x_;
  std::vector<array> splits_;
};

void init_array(py::module_& m) {
  // Set Python print formatting options
  mlx::core::global_formatter.capitalize_bool = true;

  // Types
  py::class_<Dtype>(
      m,
      "Dtype",
      R"pbdoc(
      An object to hold the type of a :class:`array`.

      See the :ref:`list of types <data_types>` for more details
      on available data types.
      )pbdoc")
      .def_readonly(
          "size", &Dtype::size, R"pbdoc(Size of the type in bytes.)pbdoc")
      .def(
          "__repr__",
          [](const Dtype& t) {
            std::ostringstream os;
            os << "mlx.core.";
            os << t;
            return os.str();
          })
      .def("__eq__", [](const Dtype& t1, const Dtype& t2) { return t1 == t2; })
      .def("__hash__", [](const Dtype& t) {
        return static_cast<int64_t>(t.val);
      });
  m.attr("bool_") = py::cast(bool_);
  m.attr("uint8") = py::cast(uint8);
  m.attr("uint16") = py::cast(uint16);
  m.attr("uint32") = py::cast(uint32);
  m.attr("uint64") = py::cast(uint64);
  m.attr("int8") = py::cast(int8);
  m.attr("int16") = py::cast(int16);
  m.attr("int32") = py::cast(int32);
  m.attr("int64") = py::cast(int64);
  m.attr("float16") = py::cast(float16);
  m.attr("float32") = py::cast(float32);
  m.attr("bfloat16") = py::cast(bfloat16);
  m.attr("complex64") = py::cast(complex64);

  auto array_at_class = py::class_<ArrayAt>(
      m,
      "_ArrayAt",
      R"pbdoc(
      A helper object to apply updates at specific indices.
      )pbdoc");

  auto array_iterator_class = py::class_<ArrayPythonIterator>(
      m,
      "_ArrayIterator",
      R"pbdoc(
      A helper object to iterate over the 1st dimension of an array.
      )pbdoc");

  auto array_class = py::class_<array>(
      m,
      "array",
      R"pbdoc(An N-dimensional array object.)pbdoc",
      py::buffer_protocol());

  {
    py::options options;
    options.disable_function_signatures();

    array_class.def(
        py::init([](array_init_type v, std::optional<Dtype> t) {
          return create_array(v, t);
        }),
        "val"_a,
        "dtype"_a = std::nullopt,
        R"pbdoc(
            __init__(self: array, val: Union[scalar, list, tuple, numpy.ndarray, array], dtype: Optional[Dtype] = None)
          )pbdoc");
  }

  array_at_class
      .def(
          py::init([](const array& x) { return ArrayAt(x); }),
          "x"_a,
          R"pbdoc(
          __init__(self, x: array)
        )pbdoc")
      .def("__getitem__", &ArrayAt::set_indices, "indices"_a)
      .def("add", &ArrayAt::add, "value"_a)
      .def("subtract", &ArrayAt::subtract, "value"_a)
      .def("multiply", &ArrayAt::multiply, "value"_a)
      .def("divide", &ArrayAt::divide, "value"_a)
      .def("maximum", &ArrayAt::maximum, "value"_a)
      .def("minimum", &ArrayAt::minimum, "value"_a);

  array_iterator_class
      .def(
          py::init([](const array& x) { return ArrayPythonIterator(x); }),
          "x"_a,
          R"pbdoc(
          __init__(self, x: array)
        )pbdoc")
      .def("__next__", &ArrayPythonIterator::next)
      .def("__iter__", [](const ArrayPythonIterator& it) { return it; });

  array_class
      .def_buffer([](array& a) {
        // Eval if not already evaled
        if (!a.is_evaled()) {
          py::gil_scoped_release nogil;
          a.eval();
        }
        return pybind11::buffer_info(
            a.data<void>(),
            a.itemsize(),
            buffer_format(a).value_or("B"), // we use "B" because pybind uses a
                                            // std::string which can't be null
            a.ndim(),
            a.shape(),
            buffer_strides(a));
      })
      .def_property_readonly(
          "size", &array::size, R"pbdoc(Number of elements in the array.)pbdoc")
      .def_property_readonly(
          "ndim", &array::ndim, R"pbdoc(The array's dimension.)pbdoc")
      .def_property_readonly(
          "itemsize",
          &array::itemsize,
          R"pbdoc(The size of the array's datatype in bytes.)pbdoc")
      .def_property_readonly(
          "nbytes",
          &array::nbytes,
          R"pbdoc(The number of bytes in the array.)pbdoc")
      .def_property_readonly(
          "shape",
          [](const array& a) { return py::tuple(py::cast(a.shape())); },
          R"pbdoc(
          The shape of the array as a Python list.

          Returns:
            tuple(int): A tuple containing the sizes of each dimension.
        )pbdoc")
      .def_property_readonly(
          "dtype",
          &array::dtype,
          R"pbdoc(
            The array's :class:`Dtype`.
          )pbdoc")
      .def(
          "item",
          &to_scalar,
          R"pbdoc(
            Access the value of a scalar array.

            Returns:
                Standard Python scalar.
          )pbdoc")
      .def(
          "tolist",
          &tolist,
          R"pbdoc(
            Convert the array to a Python :class:`list`.

            Returns:
                list: The Python list.

                If the array is a scalar then a standard Python scalar is returned.

                If the array has more than one dimension then the result is a nested
                list of lists.

                The value type of the list corresponding to the last dimension is either
                ``bool``, ``int`` or ``float`` depending on the ``dtype`` of the array.
          )pbdoc")
      .def(
          "astype",
          &astype,
          "dtype"_a,
          "stream"_a = none,
          R"pbdoc(
            Cast the array to a specified type.

            Args:
                dtype (Dtype): Type to which the array is cast.
                stream (Stream): Stream (or device) for the operation.

            Returns:
                array: The array with type ``dtype``.
          )pbdoc")
      .def("__getitem__", mlx_get_item)
      .def("__setitem__", mlx_set_item)
      .def_property_readonly(
          "at",
          [](const array& a) { return ArrayAt(a); },
          R"pbdoc(
            Used to apply updates at the given indices.

            .. note::

               Python in place updates for all array frameworks map to
               assignment. For instance ``x[idx] += y`` maps to ``x[idx] =
               x[idx] + y``. As a result, assigning to the same index ignores
               all but one updates. Using ``x.at[idx].add(y)`` will correctly
               apply all the updates to all indices.

            .. list-table::
               :header-rows: 1

               * - array.at syntax
                 - In-place syntax
               * - ``x = x.at[idx].add(y)``
                 - ``x[idx] += y``
               * - ``x = x.at[idx].subtract(y)``
                 - ``x[idx] -= y``
               * - ``x = x.at[idx].multiply(y)``
                 - ``x[idx] *= y``
               * - ``x = x.at[idx].divide(y)``
                 - ``x[idx] /= y``
               * - ``x = x.at[idx].maximum(y)``
                 - ``x[idx] = mx.maximum(x[idx], y)``
               * - ``x = x.at[idx].minimum(y)``
                - ``x[idx] = mx.minimum(x[idx], y)``
          )pbdoc")
      .def(
          "__len__",
          [](const array& a) {
            if (a.ndim() == 0) {
              throw py::type_error("len() 0-dimensional array.");
            }
            return a.shape(0);
          })
      .def("__iter__", [](const array& a) { return ArrayPythonIterator(a); })
      .def(
          "__add__",
          [](const array& a, const ScalarOrArray v) {
            return add(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__iadd__",
          [](array& a, const ScalarOrArray v) {
            a.overwrite_descriptor(add(a, to_array(v, a.dtype())));
            return a;
          },
          "other"_a)
      .def(
          "__radd__",
          [](const array& a, const ScalarOrArray v) {
            return add(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__sub__",
          [](const array& a, const ScalarOrArray v) {
            return subtract(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__isub__",
          [](array& a, const ScalarOrArray v) {
            a.overwrite_descriptor(subtract(a, to_array(v, a.dtype())));
            return a;
          },
          "other"_a)
      .def(
          "__rsub__",
          [](const array& a, const ScalarOrArray v) {
            return subtract(to_array(v, a.dtype()), a);
          },
          "other"_a)
      .def(
          "__mul__",
          [](const array& a, const ScalarOrArray v) {
            return multiply(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__imul__",
          [](array& a, const ScalarOrArray v) {
            a.overwrite_descriptor(multiply(a, to_array(v, a.dtype())));
            return a;
          },
          "other"_a)
      .def(
          "__rmul__",
          [](const array& a, const ScalarOrArray v) {
            return multiply(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__truediv__",
          [](const array& a, const ScalarOrArray v) {
            return divide(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__itruediv__",
          [](array& a, const ScalarOrArray v) {
            if (!is_floating_point(a.dtype())) {
              throw std::invalid_argument(
                  "In place division cannot cast to non-floating point type.");
            }
            a.overwrite_descriptor(divide(a, to_array(v, a.dtype())));
            return a;
          },
          "other"_a)
      .def(
          "__rtruediv__",
          [](const array& a, const ScalarOrArray v) {
            return divide(to_array(v, a.dtype()), a);
          },
          "other"_a)
      .def(
          "__div__",
          [](const array& a, const ScalarOrArray v) {
            return divide(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__rdiv__",
          [](const array& a, const ScalarOrArray v) {
            return divide(to_array(v, a.dtype()), a);
          },
          "other"_a)
      .def(
          "__floordiv__",
          [](const array& a, const ScalarOrArray v) {
            return floor_divide(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__ifloordiv__",
          [](array& a, const ScalarOrArray v) {
            a.overwrite_descriptor(floor_divide(a, to_array(v, a.dtype())));
            return a;
          },
          "other"_a)
      .def(
          "__rfloordiv__",
          [](const array& a, const ScalarOrArray v) {
            auto b = to_array(v, a.dtype());
            return floor_divide(b, a);
          },
          "other"_a)
      .def(
          "__mod__",
          [](const array& a, const ScalarOrArray v) {
            return remainder(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__imod__",
          [](array& a, const ScalarOrArray v) {
            a.overwrite_descriptor(remainder(a, to_array(v, a.dtype())));
            return a;
          },
          "other"_a)
      .def(
          "__rmod__",
          [](const array& a, const ScalarOrArray v) {
            return remainder(to_array(v, a.dtype()), a);
          },
          "other"_a)
      .def(
          "__eq__",
          [](const array& a, const ScalarOrArray v) {
            return equal(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__lt__",
          [](const array& a, const ScalarOrArray v) {
            return less(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__le__",
          [](const array& a, const ScalarOrArray v) {
            return less_equal(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__gt__",
          [](const array& a, const ScalarOrArray v) {
            return greater(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__ge__",
          [](const array& a, const ScalarOrArray v) {
            return greater_equal(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__ne__",
          [](const array& a, const ScalarOrArray v) {
            return not_equal(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def("__neg__", [](const array& a) { return -a; })
      .def("__bool__", [](array& a) { return py::bool_(to_scalar(a)); })
      .def(
          "__repr__",
          [](array& a) {
            if (!a.is_evaled()) {
              py::gil_scoped_release nogil;
              a.eval();
            }
            std::ostringstream os;
            os << a;
            return os.str();
          })
      .def(
          "__matmul__",
          [](const array& a, array& other) { return matmul(a, other); },
          "other"_a)
      .def(
          "__imatmul__",
          [](array& a, array& other) {
            a.overwrite_descriptor(matmul(a, other));
            return a;
          },
          "other"_a)
      .def(
          "__pow__",
          [](const array& a, const ScalarOrArray v) {
            return power(a, to_array(v, a.dtype()));
          },
          "other"_a)
      .def(
          "__ipow__",
          [](array& a, const ScalarOrArray v) {
            a.overwrite_descriptor(power(a, to_array(v, a.dtype())));
            return a;
          },
          "other"_a)
      .def(
          "__invert__",
          [](const array& a) {
            if (is_floating_point(a.dtype())) {
              throw std::invalid_argument(
                  "Floating point types not allowed with or bitwise inversion.");
            }
            if (a.dtype() != bool_) {
              throw std::invalid_argument(
                  "Bitwise inversion not yet supported for integer types.");
            }
            return logical_not(a);
          })
      .def(
          "__and__",
          [](const array& a, const ScalarOrArray v) {
            auto b = to_array(v, a.dtype());
            if (is_floating_point(a.dtype()) || is_floating_point(b.dtype())) {
              throw std::invalid_argument(
                  "Floating point types not allowed with bitwise and.");
            }
            if (a.dtype() != bool_ && b.dtype() != bool_) {
              throw std::invalid_argument(
                  "Bitwise and not yet supported for integer types.");
            }
            return logical_and(a, b);
          },
          "other"_a)
      .def(
          "__iand__",
          [](array& a, const ScalarOrArray v) {
            auto b = to_array(v, a.dtype());
            if (is_floating_point(a.dtype()) || is_floating_point(b.dtype())) {
              throw std::invalid_argument(
                  "Floating point types not allowed with bitwise and.");
            }
            if (a.dtype() != bool_ && b.dtype() != bool_) {
              throw std::invalid_argument(
                  "Bitwise and not yet supported for integer types.");
            }
            a.overwrite_descriptor(logical_and(a, b));
            return a;
          },
          "other"_a)
      .def(
          "__or__",
          [](const array& a, const ScalarOrArray v) {
            auto b = to_array(v, a.dtype());
            if (is_floating_point(a.dtype()) || is_floating_point(b.dtype())) {
              throw std::invalid_argument(
                  "Floating point types not allowed with or bitwise or.");
            }
            if (a.dtype() != bool_ && b.dtype() != bool_) {
              throw std::invalid_argument(
                  "Bitwise or not yet supported for integer types.");
            }
            return logical_or(a, b);
          },
          "other"_a)
      .def(
          "__ior__",
          [](array& a, const ScalarOrArray v) {
            auto b = to_array(v, a.dtype());
            if (is_floating_point(a.dtype()) || is_floating_point(b.dtype())) {
              throw std::invalid_argument(
                  "Floating point types not allowed with or bitwise or.");
            }
            if (a.dtype() != bool_ && b.dtype() != bool_) {
              throw std::invalid_argument(
                  "Bitwise or not yet supported for integer types.");
            }
            a.overwrite_descriptor(logical_or(a, b));
            return a;
          },
          "other"_a)
      .def(
          "flatten",
          [](const array& a,
             int start_axis,
             int end_axis,
             const StreamOrDevice& s) {
            return flatten(a, start_axis, end_axis);
          },
          "start_axis"_a = 0,
          "end_axis"_a = -1,
          py::kw_only(),
          "stream"_a = none,
          R"pbdoc(
            See :func:`flatten`.
          )pbdoc")
      .def(
          "reshape",
          [](const array& a, py::args shape, StreamOrDevice s) {
            if (shape.size() == 1) {
              py::object arg = shape[0];
              if (!py::isinstance<py::int_>(arg)) {
                return reshape(a, py::cast<std::vector<int>>(arg), s);
              }
            }
            return reshape(a, py::cast<std::vector<int>>(shape), s);
          },
          py::kw_only(),
          "stream"_a = none,
          R"pbdoc(
            Equivalent to :func:`reshape` but the shape can be passed either as a
            tuple or as separate arguments.

            See :func:`reshape` for full documentation.
          )pbdoc")
      .def(
          "squeeze",
          [](const array& a, const IntOrVec& v, const StreamOrDevice& s) {
            if (std::holds_alternative<std::monostate>(v)) {
              return squeeze(a, s);
            } else if (auto pv = std::get_if<int>(&v); pv) {
              return squeeze(a, *pv, s);
            } else {
              return squeeze(a, std::get<std::vector<int>>(v), s);
            }
          },
          "axis"_a = none,
          py::kw_only(),
          "stream"_a = none,
          R"pbdoc(
            See :func:`squeeze`.
          )pbdoc")
      .def(
          "abs",
          &mlx::core::abs,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`abs`.")
      .def(
          "square",
          &square,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`square`.")
      .def(
          "sqrt",
          &mlx::core::sqrt,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`sqrt`.")
      .def(
          "rsqrt",
          &rsqrt,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`rsqrt`.")
      .def(
          "reciprocal",
          &reciprocal,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`reciprocal`.")
      .def(
          "exp",
          &mlx::core::exp,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`exp`.")
      .def(
          "log",
          &mlx::core::log,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`log`.")
      .def(
          "log2",
          &mlx::core::log2,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`log2`.")
      .def(
          "log10",
          &mlx::core::log10,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`log10`.")
      .def(
          "sin",
          &mlx::core::sin,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`sin`.")
      .def(
          "cos",
          &mlx::core::cos,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`cos`.")
      .def(
          "log1p",
          &mlx::core::log1p,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`log1p`.")
      .def(
          "all",
          [](const array& a,
             const IntOrVec& axis,
             bool keepdims,
             StreamOrDevice s) {
            return all(a, get_reduce_axes(axis, a.ndim()), keepdims, s);
          },
          "axis"_a = none,
          "keepdims"_a = false,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`all`.")
      .def(
          "any",
          [](const array& a,
             const IntOrVec& axis,
             bool keepdims,
             StreamOrDevice s) {
            return any(a, get_reduce_axes(axis, a.ndim()), keepdims, s);
          },
          "axis"_a = none,
          "keepdims"_a = false,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`any`.")
      .def(
          "moveaxis",
          &moveaxis,
          "source"_a,
          "destination"_a,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`moveaxis`.")
      .def(
          "swapaxes",
          &swapaxes,
          "axis1"_a,
          "axis2"_a,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`swapaxes`.")
      .def(
          "transpose",
          [](const array& a, py::args axes, StreamOrDevice s) {
            if (axes.size() > 0) {
              if (axes.size() == 1) {
                py::object arg = axes[0];
                if (!py::isinstance<py::int_>(arg)) {
                  return transpose(a, py::cast<std::vector<int>>(arg), s);
                }
              }
              return transpose(a, py::cast<std::vector<int>>(axes), s);
            } else {
              return transpose(a, s);
            }
          },
          py::kw_only(),
          "stream"_a = none,
          R"pbdoc(
            Equivalent to :func:`transpose` but the axes can be passed either as
            a tuple or as separate arguments.

            See :func:`transpose` for full documentation.
          )pbdoc")
      .def_property_readonly(
          "T",
          [](const array& a) { return transpose(a); },
          "Equivalent to calling ``self.transpose()`` with no arguments.")
      .def(
          "sum",
          [](const array& a,
             const IntOrVec& axis,
             bool keepdims,
             StreamOrDevice s) {
            return sum(a, get_reduce_axes(axis, a.ndim()), keepdims, s);
          },
          "axis"_a = none,
          "keepdims"_a = false,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`sum`.")
      .def(
          "prod",
          [](const array& a,
             const IntOrVec& axis,
             bool keepdims,
             StreamOrDevice s) {
            return prod(a, get_reduce_axes(axis, a.ndim()), keepdims, s);
          },
          "axis"_a = none,
          "keepdims"_a = false,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`prod`.")
      .def(
          "min",
          [](const array& a,
             const IntOrVec& axis,
             bool keepdims,
             StreamOrDevice s) {
            return min(a, get_reduce_axes(axis, a.ndim()), keepdims, s);
          },
          "axis"_a = none,
          "keepdims"_a = false,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`min`.")
      .def(
          "max",
          [](const array& a,
             const IntOrVec& axis,
             bool keepdims,
             StreamOrDevice s) {
            return max(a, get_reduce_axes(axis, a.ndim()), keepdims, s);
          },
          "axis"_a = none,
          "keepdims"_a = false,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`max`.")
      .def(
          "logsumexp",
          [](const array& a,
             const IntOrVec& axis,
             bool keepdims,
             StreamOrDevice s) {
            return logsumexp(a, get_reduce_axes(axis, a.ndim()), keepdims, s);
          },
          "axis"_a = none,
          "keepdims"_a = false,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`logsumexp`.")
      .def(
          "mean",
          [](const array& a,
             const IntOrVec& axis,
             bool keepdims,
             StreamOrDevice s) {
            return mean(a, get_reduce_axes(axis, a.ndim()), keepdims, s);
          },
          "axis"_a = none,
          "keepdims"_a = false,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`mean`.")
      .def(
          "var",
          [](const array& a,
             const IntOrVec& axis,
             bool keepdims,
             int ddof,
             StreamOrDevice s) {
            return var(a, get_reduce_axes(axis, a.ndim()), keepdims, ddof, s);
          },
          "axis"_a = none,
          "keepdims"_a = false,
          "ddof"_a = 0,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`var`.")
      .def(
          "split",
          [](const array& a,
             const std::variant<int, std::vector<int>>& indices_or_sections,
             int axis,
             StreamOrDevice s) {
            if (auto pv = std::get_if<int>(&indices_or_sections); pv) {
              return split(a, *pv, axis, s);
            } else {
              return split(
                  a, std::get<std::vector<int>>(indices_or_sections), axis, s);
            }
          },
          "indices_or_sections"_a,
          "axis"_a = 0,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`split`.")
      .def(
          "argmin",
          [](const array& a,
             std::optional<int> axis,
             bool keepdims,
             StreamOrDevice s) {
            if (axis) {
              return argmin(a, *axis, keepdims, s);
            } else {
              return argmin(a, keepdims, s);
            }
          },
          "axis"_a = std::nullopt,
          "keepdims"_a = false,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`argmin`.")
      .def(
          "argmax",
          [](const array& a,
             std::optional<int> axis,
             bool keepdims,
             StreamOrDevice s) {
            if (axis) {
              return argmax(a, *axis, keepdims, s);
            } else {
              return argmax(a, keepdims, s);
            }
          },
          "axis"_a = none,
          "keepdims"_a = false,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`argmax`.")
      .def(
          "cumsum",
          [](const array& a,
             std::optional<int> axis,
             bool reverse,
             bool inclusive,
             StreamOrDevice s) {
            if (axis) {
              return cumsum(a, *axis, reverse, inclusive, s);
            } else {
              // TODO: Implement that in the C++ API as well. See concatenate
              // above.
              return cumsum(reshape(a, {-1}, s), 0, reverse, inclusive, s);
            }
          },
          "axis"_a = none,
          py::kw_only(),
          "reverse"_a = false,
          "inclusive"_a = true,
          "stream"_a = none,
          "See :func:`cumsum`.")
      .def(
          "cumprod",
          [](const array& a,
             std::optional<int> axis,
             bool reverse,
             bool inclusive,
             StreamOrDevice s) {
            if (axis) {
              return cumprod(a, *axis, reverse, inclusive, s);
            } else {
              // TODO: Implement that in the C++ API as well. See concatenate
              // above.
              return cumprod(reshape(a, {-1}, s), 0, reverse, inclusive, s);
            }
          },
          "axis"_a = none,
          py::kw_only(),
          "reverse"_a = false,
          "inclusive"_a = true,
          "stream"_a = none,
          "See :func:`cumprod`.")
      .def(
          "cummax",
          [](const array& a,
             std::optional<int> axis,
             bool reverse,
             bool inclusive,
             StreamOrDevice s) {
            if (axis) {
              return cummax(a, *axis, reverse, inclusive, s);
            } else {
              // TODO: Implement that in the C++ API as well. See concatenate
              // above.
              return cummax(reshape(a, {-1}, s), 0, reverse, inclusive, s);
            }
          },
          "axis"_a = none,
          py::kw_only(),
          "reverse"_a = false,
          "inclusive"_a = true,
          "stream"_a = none,
          "See :func:`cummax`.")
      .def(
          "cummin",
          [](const array& a,
             std::optional<int> axis,
             bool reverse,
             bool inclusive,
             StreamOrDevice s) {
            if (axis) {
              return cummin(a, *axis, reverse, inclusive, s);
            } else {
              // TODO: Implement that in the C++ API as well. See concatenate
              // above.
              return cummin(reshape(a, {-1}, s), 0, reverse, inclusive, s);
            }
          },
          "axis"_a = none,
          py::kw_only(),
          "reverse"_a = false,
          "inclusive"_a = true,
          "stream"_a = none,
          "See :func:`cummin`.")
      .def(
          "round",
          [](const array& a, int decimals, StreamOrDevice s) {
            return round(a, decimals, s);
          },
          py::pos_only(),
          "decimals"_a = 0,
          py::kw_only(),
          "stream"_a = none,
          "See :func:`round`.")
      .def(
          "diagonal",
          [](const array& a,
             int offset,
             int axis1,
             int axis2,
             StreamOrDevice s) { return diagonal(a, offset, axis1, axis2, s); },
          "offset"_a = 0,
          "axis1"_a = 0,
          "axis2"_a = 1,
          "stream"_a = none,
          "See :func:`diagonal`.")
      .def(
          "diag",
          [](const array& a, int k, StreamOrDevice s) { return diag(a, k, s); },
          "k"_a = 0,
          py::kw_only(),
          "stream"_a = none,
          R"pbdoc(
            Extract a diagonal or construct a diagonal matrix.
        )pbdoc");
}