mlx/docs/build/doctrees/usage/numpy.doctree

<EFBFBD><05>C8<00>sphinx.addnodes<65><73>document<6E><74><EFBFBD>)<29><>}<7D>(<28>	rawsource<63><65><00><>children<65>]<5D>(<28>docutils.nodes<65><73>target<65><74><EFBFBD>)<29><>}<7D>(h<05>
.. _numpy:<3A>h]<5D><>
attributes<EFBFBD>}<7D>(<28>ids<64>]<5D><>classes<65>]<5D><>names<65>]<5D><>dupnames<65>]<5D><>backrefs<66>]<5D><>refid<69><64>numpy<70>u<EFBFBD>tagname<6D>h
<EFBFBD>line<6E>K
<01>parent<6E>h<03>	_document<6E>h<03>source<63><65>4/Users/awnihannun/repos/mlx/docs/src/usage/numpy.rst<73>ubh	<09>section<6F><6E><EFBFBD>)<29><>}<7D>(hhh]<5D>(h	<09>title<6C><65><EFBFBD>)<29><>}<7D>(h<05>(Conversion to NumPy and Other Frameworks<6B>h]<5D>h	<09>Text<78><74><EFBFBD><EFBFBD>(Conversion to NumPy and Other Frameworks<6B><73><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h+h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh)h h&h!hh"h#hKubh	<09>	paragraph<70><68><EFBFBD>)<29><>}<7D>(h<05><>MLX array implements the `Python Buffer Protocol <https://docs.python.org/3/c-api/buffer.html>`_.
Let's convert an array to NumPy and back.<2E>h]<5D>(h0<68>MLX array implements the <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h=h!hh"NhNubh	<09>	reference<63><65><EFBFBD>)<29><>}<7D>(h<05>G`Python Buffer Protocol <https://docs.python.org/3/c-api/buffer.html>`_<>h]<5D>h0<68>Python Buffer Protocol<6F><6C><EFBFBD><EFBFBD><EFBFBD>}<7D>(h hGh!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D><>name<6D><65>Python Buffer Protocol<6F><6C>refuri<72><69>+https://docs.python.org/3/c-api/buffer.html<6D>uhhEh h=ubh)<29><>}<7D>(h<05>. <https://docs.python.org/3/c-api/buffer.html><3E>h]<5D>h}<7D>(h]<5D><>python-buffer-protocol<6F>ah]<5D>h]<5D><>python buffer protocol<6F>ah]<5D>h]<5D><>refuri<72>hXuhh
<EFBFBD>
referenced<EFBFBD>K
h h=ubh0<68>-.
Let’s convert an array to NumPy and back.<2E><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h=h!hh"NhNubeh}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hKh h&h!hubh	<09>
literal_block<63><6B><EFBFBD>)<29><>}<7D>(h<05>rimport mlx.core as mx
import numpy as np

a = mx.arange(3)
b = np.array(a) # copy of a
c = mx.array(b) # copy of b<>h]<5D>h0<68>rimport mlx.core as mx
import numpy as np

a = mx.arange(3)
b = np.array(a) # copy of a
c = mx.array(b) # copy of b<><62><EFBFBD><EFBFBD><EFBFBD>}<7D>h hssbah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D><>	xml:space<63><65>preserve<76><65>force<63><65><EFBFBD>language<67><65>python<6F><6E>highlight_args<67>}<7D>uhhqh"h#hK	h h&h!hubh	<09>note<74><65><EFBFBD>)<29><>}<7D>(hX
Since NumPy does not support ``bfloat16`` arrays, you will need to convert to ``float16`` or ``float32`` first:
``np.array(a.astype(mx.float32))``.
Otherwise, you will receive an error like: ``Item size 2 for PEP 3118 buffer format string does not match the dtype V item size 0.``<60>h]<5D>h<)<29><>}<7D>(hX
Since NumPy does not support ``bfloat16`` arrays, you will need to convert to ``float16`` or ``float32`` first:
``np.array(a.astype(mx.float32))``.
Otherwise, you will receive an error like: ``Item size 2 for PEP 3118 buffer format string does not match the dtype V item size 0.``<60>h]<5D>(h0<68>Since NumPy does not support <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h<>h!hh"NhNubh	<09>literal<61><6C><EFBFBD>)<29><>}<7D>(h<05>``bfloat16``<60>h]<5D>h0<68>bfloat16<31><36><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h<>h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h h<>ubh0<68>% arrays, you will need to convert to <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h<>h!hh"NhNubh<62>)<29><>}<7D>(h<05>``float16``<60>h]<5D>h0<68>float16<31><36><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h<>h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h h<>ubh0<68> or <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h<>h!hh"NhNubh<62>)<29><>}<7D>(h<05>``float32``<60>h]<5D>h0<68>float32<33><32><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h<>h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h h<>ubh0<68> first:
<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h<>h!hh"NhNubh<62>)<29><>}<7D>(h<05>"``np.array(a.astype(mx.float32))``<60>h]<5D>h0<68>np.array(a.astype(mx.float32))<29><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h<>h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h h<>ubh0<68>-.
Otherwise, you will receive an error like: <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h<>h!hh"NhNubh<62>)<29><>}<7D>(h<05>Y``Item size 2 for PEP 3118 buffer format string does not match the dtype V item size 0.``<60>h]<5D>h0<68>UItem size 2 for PEP 3118 buffer format string does not match the dtype V item size 0.<2E><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h<>h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h h<>ubeh}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hKh h<>ubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h h&h!hh"h#hNubh<)<29><>}<7D>(h<05>^By default, NumPy copies data to a new array. This can be prevented by creating an array view:<3A>h]<5D>h0<68>^By default, NumPy copies data to a new array. This can be prevented by creating an array view:<3A><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h h<>h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hKh h&h!hubhr)<29><>}<7D>(h<05>za = mx.arange(3)
a_view = np.array(a, copy=False)
print(a_view.flags.owndata) # False
a_view[0] = 1
print(a[0].item()) # 1<>h]<5D>h0<68>za = mx.arange(3)
a_view = np.array(a, copy=False)
print(a_view.flags.owndata) # False
a_view[0] = 1
print(a[0].item()) # 1<><31><EFBFBD><EFBFBD><EFBFBD>}<7D>h j
sbah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>h<EFBFBD>h<EFBFBD>h<EFBFBD><68>h<EFBFBD><68>python<6F>h<EFBFBD>}<7D>uhhqh"h#hKh h&h!hubh<)<29><>}<7D>(h<05><>A NumPy array view is a normal NumPy array, except that it does not own its memory.
This means writing to the view is reflected in the original array.<2E>h]<5D>h0<68><30>A NumPy array view is a normal NumPy array, except that it does not own its memory.
This means writing to the view is reflected in the original array.<2E><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j
h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hK"h h&h!hubh<)<29><>}<7D>(h<05><>While this is quite powerful to prevent copying arrays, it should be noted that external changes to the memory of arrays cannot be reflected in gradients.<2E>h]<5D>h0<68><30>While this is quite powerful to prevent copying arrays, it should be noted that external changes to the memory of arrays cannot be reflected in gradients.<2E><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j&
h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hK%h h&h!hubh<)<29><>}<7D>(h<05>%Let's demonstrate this in an example:<3A>h]<5D>h0<68>'Let’s demonstrate this in an example:<3A><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j4
h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hK'h h&h!hubhr)<29><>}<7D>(h<05><>def f(x):
    x_view = np.array(x, copy=False)
    x_view[:] *= x_view # modify memory without telling mx
    return x.sum()

x = mx.array([3.0])
y, df = mx.value_and_grad(f)(x)
print("f(x) = x² =", y.item()) # 9.0
print("f'(x) = 2x !=", df.item()) # 1.0<EFBFBD>h]<5D>h0<68><30>def f(x):
    x_view = np.array(x, copy=False)
    x_view[:] *= x_view # modify memory without telling mx
    return x.sum()

x = mx.array([3.0])
y, df = mx.value_and_grad(f)(x)
print("f(x) = x² =", y.item()) # 9.0
print("f'(x) = 2x !=", df.item()) # 1.0<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>}<7D>h jB
sbah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>h<EFBFBD>h<EFBFBD>h<EFBFBD><68>h<EFBFBD><68>python<6F>h<EFBFBD>}<7D>uhhqh"h#hK)h h&h!hubh<)<29><>}<7D>(hXQThe function ``f`` indirectly modifies the array ``x`` through a memory view.
However, this modification is not reflected in the gradient, as seen in the last line outputting ``1.0``,
representing the gradient of the sum operation alone.
The squaring of ``x`` occurs externally to MLX, meaning that no gradient is incorporated.
It's important to note that a similar issue arises during array conversion and copying.
For instance, a function defined as ``mx.array(np.array(x)**2).sum()`` would also result in an incorrect gradient,
even though no in-place operations on MLX memory are executed.<2E>h]<5D>(h0<68>
The function <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jR
h!hh"NhNubh<62>)<29><>}<7D>(h<05>``f``<60>h]<5D>h0<68>
f<><66><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jZ
h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h jR
ubh0<68> indirectly modifies the array <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jR
h!hh"NhNubh<62>)<29><>}<7D>(h<05>``x``<60>h]<5D>h0<68>
x<><78><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jl
h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h jR
ubh0<68>y through a memory view.
However, this modification is not reflected in the gradient, as seen in the last line outputting <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jR
h!hh"NhNubh<62>)<29><>}<7D>(h<05>``1.0``<60>h]<5D>h0<68>1.0<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j~
h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h jR
ubh0<68>H,
representing the gradient of the sum operation alone.
The squaring of <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jR
h!hh"NhNubh<62>)<29><>}<7D>(h<05>``x``<60>h]<5D>h0<68>
x<><78><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j<>
h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h jR
ubh0<68><30> occurs externally to MLX, meaning that no gradient is incorporated.
It’s important to note that a similar issue arises during array conversion and copying.
For instance, a function defined as <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jR
h!hh"NhNubh<62>)<29><>}<7D>(h<05>"``mx.array(np.array(x)**2).sum()``<60>h]<5D>h0<68>mx.array(np.array(x)**2).sum()<29><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j<>
h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h jR
ubh0<68>k would also result in an incorrect gradient,
even though no in-place operations on MLX memory are executed.<2E><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jR
h!hh"NhNubeh}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hK6h h&h!hubh%)<29><>}<7D>(hhh]<5D>(h*)<29><>}<7D>(h<05>PyTorch<63>h]<5D>h0<68>PyTorch<63><68><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j<>
h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh)h j<>
h!hh"h#hK?ubh	<09>warning<6E><67><EFBFBD>)<29><>}<7D>(h<05><>PyTorch Support for :obj:`memoryview` is experimental and can break for
multi-dimensional arrays. Casting to NumPy first is advised for now.<2E>h]<5D>h<)<29><>}<7D>(h<05><>PyTorch Support for :obj:`memoryview` is experimental and can break for
multi-dimensional arrays. Casting to NumPy first is advised for now.<2E>h]<5D>(h0<68>PyTorch Support for <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j<>
h!hh"NhNubh<00>pending_xref<65><66><EFBFBD>)<29><>}<7D>(h<05>:obj:`memoryview`<60>h]<5D>h<EFBFBD>)<29><>}<7D>(hj<>
h]<5D>h0<68>
memoryview<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j<>
h!hh"NhNubah}<7D>(h]<5D>h]<5D>(<28>xref<65><66>py<70><79>py-obj<62>eh]<5D>h]<5D>h]<5D>uhh<>h j<>
ubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D><>refdoc<6F><63>usage/numpy<70><79>	refdomain<69>j<EFBFBD>
<00>reftype<70><65>obj<62><6A>refexplicit<69><74><EFBFBD>refwarn<72><6E><EFBFBD>	py:module<6C>N<EFBFBD>py:class<73>N<EFBFBD>	reftarget<65><74>
memoryview<EFBFBD>uhj<>
h"h#hKCh j<>
ubh0<68>g is experimental and can break for
multi-dimensional arrays. Casting to NumPy first is advised for now.<2E><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j<>
h!hh"NhNubeh}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hKCh j<>
ubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhj<>
h j<>
h!hh"h#hNubh<)<29><>}<7D>(h<05>TPyTorch supports the buffer protocol, but it requires an explicit :obj:`memoryview`.<2E>h]<5D>(h0<68>BPyTorch supports the buffer protocol, but it requires an explicit <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jh!hh"NhNubj<62>
)<29><>}<7D>(h<05>:obj:`memoryview`<60>h]<5D>h<EFBFBD>)<29><>}<7D>(hjh]<5D>h0<68>
memoryview<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jh!hh"NhNubah}<7D>(h]<5D>h]<5D>(j<>
<00>py<70><79>py-obj<62>eh]<5D>h]<5D>h]<5D>uhh<>h jubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D><>refdoc<6F>j<EFBFBD>
<00>	refdomain<69>j&<00>reftype<70><65>obj<62><6A>refexplicit<69><74><EFBFBD>refwarn<72><6E>j<EFBFBD>
Nj<4E>
Nj<4E>
<00>
memoryview<EFBFBD>uhj<>
h"h#hKFh jubh0<68>
.<2E><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jh!hh"NhNubeh}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hKFh j<>
h!hubhr)<29><>}<7D>(h<05>limport mlx.core as mx
import torch

a = mx.arange(3)
b = torch.tensor(memoryview(a))
c = mx.array(b.numpy())<29>h]<5D>h0<68>limport mlx.core as mx
import torch

a = mx.arange(3)
b = torch.tensor(memoryview(a))
c = mx.array(b.numpy())<29><><EFBFBD><EFBFBD><EFBFBD>}<7D>h jBsbah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>h<EFBFBD>h<EFBFBD>h<EFBFBD><68>h<EFBFBD><68>python<6F>h<EFBFBD>}<7D>uhhqh"h#hKHh j<>
h!hubh<)<29><>}<7D>(h<05>kConversion from PyTorch tensors back to arrays must be done via intermediate NumPy arrays with ``numpy()``.<2E>h]<5D>(h0<68>_Conversion from PyTorch tensors back to arrays must be done via intermediate NumPy arrays with <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jRh!hh"NhNubh<62>)<29><>}<7D>(h<05>``numpy()``<60>h]<5D>h0<68>numpy()<29><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jZh!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh<>h jRubh0<68>
.<2E><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h jRh!hh"NhNubeh}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hKQh j<>
h!hubeh}<7D>(h]<5D><>pytorch<63>ah]<5D>h]<5D><>pytorch<63>ah]<5D>h]<5D>uhh$h h&h!hh"h#hK?ubh%)<29><>}<7D>(hhh]<5D>(h*)<29><>}<7D>(h<05>JAX<41>h]<5D>h0<68>JAX<41><58><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j}h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh)h jzh!hh"h#hKTubh<)<29><>}<7D>(h<05>'JAX fully supports the buffer protocol.<2E>h]<5D>h0<68>'JAX fully supports the buffer protocol.<2E><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j<>h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hKUh jzh!hubhr)<29><>}<7D>(h<05>`import mlx.core as mx
import jax.numpy as jnp

a = mx.arange(3)
b = jnp.array(a)
c = mx.array(b)<29>h]<5D>h0<68>`import mlx.core as mx
import jax.numpy as jnp

a = mx.arange(3)
b = jnp.array(a)
c = mx.array(b)<29><><EFBFBD><EFBFBD><EFBFBD>}<7D>h j<>sbah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>h<EFBFBD>h<EFBFBD>h<EFBFBD><68>h<EFBFBD><68>python<6F>h<EFBFBD>}<7D>uhhqh"h#hKWh jzh!hubeh}<7D>(h]<5D><>jax<61>ah]<5D>h]<5D><>jax<61>ah]<5D>h]<5D>uhh$h h&h!hh"h#hKTubh%)<29><>}<7D>(hhh]<5D>(h*)<29><>}<7D>(h<05>
TensorFlow<EFBFBD>h]<5D>h0<68>
TensorFlow<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j<>h!hh"NhNubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh)h j<>h!hh"h#hKaubh<)<29><>}<7D>(h<05>WTensorFlow supports the buffer protocol, but it requires an explicit :obj:`memoryview`.<2E>h]<5D>(h0<68>ETensorFlow supports the buffer protocol, but it requires an explicit <20><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j<>h!hh"NhNubj<62>
)<29><>}<7D>(h<05>:obj:`memoryview`<60>h]<5D>h<EFBFBD>)<29><>}<7D>(hj<>h]<5D>h0<68>
memoryview<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j<>h!hh"NhNubah}<7D>(h]<5D>h]<5D>(j<>
<00>py<70><79>py-obj<62>eh]<5D>h]<5D>h]<5D>uhh<>h j<>ubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D><>refdoc<6F>j<EFBFBD>
<00>	refdomain<69>j<EFBFBD><00>reftype<70><65>obj<62><6A>refexplicit<69><74><EFBFBD>refwarn<72><6E>j<EFBFBD>
Nj<4E>
Nj<4E>
<00>
memoryview<EFBFBD>uhj<>
h"h#hKch j<>ubh0<68>
.<2E><><EFBFBD><EFBFBD><EFBFBD>}<7D>(h j<>h!hh"NhNubeh}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h"h#hKch j<>h!hubhr)<29><>}<7D>(h<05>nimport mlx.core as mx
import tensorflow as tf

a = mx.arange(3)
b = tf.constant(memoryview(a))
c = mx.array(b)<29>h]<5D>h0<68>nimport mlx.core as mx
import tensorflow as tf

a = mx.arange(3)
b = tf.constant(memoryview(a))
c = mx.array(b)<29><><EFBFBD><EFBFBD><EFBFBD>}<7D>h j<>sbah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>h<EFBFBD>h<EFBFBD>h<EFBFBD><68>h<EFBFBD><68>python<6F>h<EFBFBD>}<7D>uhhqh"h#hKeh j<>h!hubeh}<7D>(h]<5D><>
tensorflow<EFBFBD>ah]<5D>h]<5D><>
tensorflow<EFBFBD>ah]<5D>h]<5D>uhh$h h&h!hh"h#hKaubeh}<7D>(h]<5D>(<28>(conversion-to-numpy-and-other-frameworks<6B>heh]<5D>h]<5D>(<28>(conversion to numpy and other frameworks<6B><73>numpy<70>eh]<5D>h]<5D>uhh$h hh!hh"h#hK<04>expect_referenced_by_name<6D>}<7D>jhs<>expect_referenced_by_id<69>}<7D>hhsubeh}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D><>source<63>h#uhh
<01>current_source<63>N<EFBFBD>current_line<6E>N<EFBFBD>settings<67><73>docutils.frontend<6E><64>Values<65><73><EFBFBD>)<29><>}<7D>(h)N<>	generator<6F>N<EFBFBD>	datestamp<6D>N<EFBFBD>source_link<6E>N<EFBFBD>
source_url<EFBFBD>N<EFBFBD>
toc_backlinks<6B><73>entry<72><79>footnote_backlinks<6B>K
<01>
sectnum_xform<72>K
<01>strip_comments<74>N<EFBFBD>strip_elements_with_classes<65>N<EFBFBD>
strip_classes<65>N<EFBFBD>report_level<65>K<02>
halt_level<EFBFBD>K<05>exit_status_level<65>K<05>debug<75>N<EFBFBD>warning_stream<61>N<EFBFBD>	traceback<63><6B><EFBFBD>input_encoding<6E><67>	utf-8-sig<69><67>input_encoding_error_handler<65><72>strict<63><74>output_encoding<6E><67>utf-8<><38>output_encoding_error_handler<65>j<<00>error_encoding<6E><67>utf-8<><38>error_encoding_error_handler<65><72>backslashreplace<63><65>
language_code<64><65>en<65><6E>record_dependencies<65>N<EFBFBD>config<69>N<EFBFBD>	id_prefix<69>h<06>auto_id_prefix<69><78>id<69><64>
dump_settings<67>N<EFBFBD>dump_internals<6C>N<EFBFBD>dump_transforms<6D>N<EFBFBD>dump_pseudo_xml<6D>N<EFBFBD>expose_internals<6C>N<EFBFBD>strict_visitor<6F>N<EFBFBD>_disable_config<69>N<EFBFBD>_source<63>h#<23>_destination<6F>N<EFBFBD>
_config_files<65>]<5D><>file_insertion_enabled<65><64><EFBFBD>raw_enabled<65>K
<01>line_length_limit<69>M'<27>pep_references<65>N<EFBFBD>pep_base_url<72><6C>https://peps.python.org/<2F><>pep_file_url_template<74><65>pep-%04d<34><64>rfc_references<65>N<EFBFBD>rfc_base_url<72><6C>&https://datatracker.ietf.org/doc/html/<2F><>	tab_width<74>K<08>trim_footnote_reference_space<63><65><EFBFBD>syntax_highlight<68><74>long<6E><67>smart_quotes<65><73><EFBFBD>smartquotes_locales<65>]<5D><>character_level_inline_markup<75><70><EFBFBD>doctitle_xform<72><6D><EFBFBD>
docinfo_xform<72>K
<01>sectsubtitle_xform<72><6D><EFBFBD>
image_loading<6E><67>link<6E><6B>embed_stylesheet<65><74><EFBFBD>cloak_email_addresses<65><73><EFBFBD>section_self_link<6E><6B><EFBFBD>env<6E>Nub<75>reporter<65>N<EFBFBD>indirect_targets<74>]<5D><>substitution_defs<66>}<7D><>substitution_names<65>}<7D><>refnames<65>}<7D><>refids<64>}<7D>h]<5D>has<61>nameids<64>}<7D>(jhjjhbh_jwjtj<>j<>j	ju<>	nametypes<65>}<7D>(j<00>j<00>hb<68>jw<00>j<EFBFBD><00>j	<00>uh}<7D>(hh&jh&h_hYjtj<>
j<>jzjj<>u<>
footnote_refs<66>}<7D><>
citation_refs<66>}<7D><>
autofootnotes<65>]<5D><>autofootnote_refs<66>]<5D><>symbol_footnotes<65>]<5D><>symbol_footnote_refs<66>]<5D><>	footnotes<65>]<5D><>	citations<6E>]<5D><>autofootnote_start<72>K
<01>symbol_footnote_start<72>K<00>
id_counter<EFBFBD><EFBFBD>collections<6E><73>Counter<65><72><EFBFBD>}<7D><><EFBFBD>R<EFBFBD><52>parse_messages<65>]<5D><>transform_messages<65>]<5D>h	<09>system_message<67><65><EFBFBD>)<29><>}<7D>(hhh]<5D>h<)<29><>}<7D>(hhh]<5D>h0<68>+Hyperlink target "numpy" is not referenced.<2E><><EFBFBD><EFBFBD><EFBFBD>}<7D>h j<>sbah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D>uhh;h j<>ubah}<7D>(h]<5D>h]<5D>h]<5D>h]<5D>h]<5D><>level<65>K
<01>type<70><65>INFO<46><4F>source<63>h#<23>line<6E>K
uhj<>uba<62>transformer<65>N<EFBFBD>include_log<6F>]<5D><>
decoration<EFBFBD>Nh!hub.