# Chex
Chex is a library of utilities for helping to write reliable JAX code.
This includes utils to help:
* Instrument your code (e.g. assertions, warnings)
* Debug (e.g. transforming `pmaps` in `vmaps` within a context manager).
* Test JAX code across many `variants` (e.g. jitted vs non-jitted).
## Installation
You can install the latest released version of Chex from PyPI via:
```sh
pip install chex
```
or you can install the latest development version from GitHub:
```sh
pip install git+https://github.com/deepmind/chex.git
```
## Modules Overview
### Dataclass ([dataclass.py](https://github.com/deepmind/chex/blob/master/chex/_src/dataclass.py))
Dataclasses are a popular construct introduced by Python 3.7 to allow to
easily specify typed data structures with minimal boilerplate code. They are
not, however, compatible with JAX and
[dm-tree](https://github.com/deepmind/tree) out of the box.
In Chex we provide a JAX-friendly dataclass implementation reusing python [dataclasses](https://docs.python.org/3/library/dataclasses.html#module-dataclasses).
Chex implementation of `dataclass` registers dataclasses as internal [_PyTree_
nodes](https://jax.readthedocs.io/en/latest/pytrees.html) to ensure
compatibility with JAX data structures.
In addition, we provide a class wrapper that exposes dataclasses as
`collections.Mapping` descendants which allows to process them
(e.g. (un-)flatten) in `dm-tree` methods as usual Python dictionaries.
See [`@mappable_dataclass`](https://github.com/deepmind/chex/blob/master/chex/_src/dataclass.py#L27)
docstring for more details.
Example:
```python
@chex.dataclass
class Parameters:
x: chex.ArrayDevice
y: chex.ArrayDevice
parameters = Parameters(
x=jnp.ones((2, 2)),
y=jnp.ones((1, 2)),
)
# Dataclasses can be treated as JAX pytrees
jax.tree_util.tree_map(lambda x: 2.0 * x, parameters)
# and as mappings by dm-tree
tree.flatten(parameters)
```
**NOTE**: Unlike standard Python 3.7 dataclasses, Chex
dataclasses cannot be constructed using positional arguments. They support
construction arguments provided in the same format as the Python dict
constructor. Dataclasses can be converted to tuples with the `from_tuple` and
`to_tuple` methods if necessary.
```python
parameters = Parameters(
jnp.ones((2, 2)),
jnp.ones((1, 2)),
)
# ValueError: Mappable dataclass constructor doesn't support positional args.
```
### Assertions ([asserts.py](https://github.com/deepmind/chex/blob/master/chex/_src/asserts.py))
One limitation of PyType annotations for JAX is that they do not support the
specification of `DeviceArray` ranks, shapes or dtypes. Chex includes a number
of functions that allow flexible and concise specification of these properties.
E.g. suppose you want to ensure that all tensors `t1`, `t2`, `t3` have the same
shape, and that tensors `t4`, `t5` have rank `2` and (`3` or `4`), respectively.
```python
chex.assert_equal_shape([t1, t2, t3])
chex.assert_rank([t4, t5], [2, {3, 4}])
```
More examples:
```python
from chex import assert_shape, assert_rank, ...
assert_shape(x, (2, 3)) # x has shape (2, 3)
assert_shape([x, y], [(), (2,3)]) # x is scalar and y has shape (2, 3)
assert_rank(x, 0) # x is scalar
assert_rank([x, y], [0, 2]) # x is scalar and y is a rank-2 array
assert_rank([x, y], {0, 2}) # x and y are scalar OR rank-2 arrays
assert_type(x, int) # x has type `int` (x can be an array)
assert_type([x, y], [int, float]) # x has type `int` and y has type `float`
assert_equal_shape([x, y, z]) # x, y, and z have equal shapes
assert_trees_all_close(tree_x, tree_y) # values and structure of trees match
assert_tree_all_finite(tree_x) # all tree_x leaves are finite
assert_devices_available(2, 'gpu') # 2 GPUs available
assert_tpu_available() # at least 1 TPU available
assert_numerical_grads(f, (x, y), j) # f^{(j)}(x, y) matches numerical grads
```
See `asserts.py`
[documentation](https://chex.readthedocs.io/en/latest/api.html#assertions) to
find all supported assertions.
If you cannot find a specific assertion, please consider making a pull request
or openning an issue on
[the bug tracker](https://github.com/deepmind/chex/issues).
#### Optional Arguments
All chex assertions support the following optional kwargs for manipulating the
emitted exception messages:
* `custom_message`: A string to include into the emitted exception messages.
* `include_default_message`: Whether to include the default Chex message into
the emitted exception messages.
* `exception_type`: An exception type to use. `AssertionError` by default.
For example, the following code:
```python
dataset = load_dataset()
params = init_params()
for i in range(num_steps):
params = update_params(params, dataset.sample())
chex.assert_tree_all_finite(params,
custom_message=f'Failed at iteration {i}.',
exception_type=ValueError)
```
will raise a `ValueError` that includes a step number when `params` get polluted
with `NaNs` or `None`s.
#### Static and Value (aka *Runtime*) Assertions
Chex divides all assertions into 2 classes: ***static*** and ***value***
assertions.
1. ***static*** assertions use anything except concrete values of tensors.
Examples: `assert_shape`, `assert_trees_all_equal_dtypes`,
`assert_max_traces`.
2. ***value*** assertions require access to tensor values, which are not
available during JAX tracing (see
[HowJAX primitives work](https://jax.readthedocs.io/en/latest/notebooks/How_JAX_primitives_work.html)),
thus such assertion need special treatment in a *jitted* code.
To enable value assertions in a jitted function, it can be decorated with
`chex.chexify()` wrapper. Example:
```python
@chex.chexify
@jax.jit
def logp1_abs_safe(x: chex.Array) -> chex.Array:
chex.assert_tree_all_finite(x)
return jnp.log(jnp.abs(x) + 1)
logp1_abs_safe(jnp.ones(2)) # OK
logp1_abs_safe(jnp.array([jnp.nan, 3])) # FAILS (in async mode)
# The error will be raised either at the next line OR at the next
# `logp1_abs_safe` call. See the docs for more detain on async mode.
logp1_abs_safe.wait_checks() # Wait for the (async) computation to complete.
```
See
[this docstring](https://chex.readthedocs.io/en/latest/api.html#chex.chexify)
for more detail on `chex.chexify()`.
#### JAX Tracing Assertions
JAX re-traces JIT'ted function every time the structure of passed arguments
changes. Often this behavior is inadvertent and leads to a significant
performance drop which is hard to debug. [@chex.assert_max_traces](https://github.com/deepmind/chex/blob/master/chex/_src/asserts.py#L44)
decorator asserts that the function is not re-traced more than `n` times during
program execution.
Global trace counter can be cleared by calling
`chex.clear_trace_counter()`. This function be used to isolate unittests relying
on `@chex.assert_max_traces`.
Examples:
```python
@jax.jit
@chex.assert_max_traces(n=1)
def fn_sum_jitted(x, y):
return x + y
fn_sum_jitted(jnp.zeros(3), jnp.zeros(3)) # tracing for the 1st time - OK
fn_sum_jitted(jnp.zeros([6, 7]), jnp.zeros([6, 7])) # AssertionError!
```
Can be used with `jax.pmap()` as well:
```python
def fn_sub(x, y):
return x - y
fn_sub_pmapped = jax.pmap(chex.assert_max_traces(fn_sub, n=10))
```
See
[HowJAX primitives work](https://jax.readthedocs.io/en/latest/notebooks/How_JAX_primitives_work.html)
section for more information about tracing.
### Warnings ([warnigns.py](https://github.com/deepmind/chex/blob/master/chex/_src/warnings.py))
In addition to hard assertions Chex also offers utilities to add common
warnings, such as specific types of deprecation warnings.
### Test variants ([variants.py](https://github.com/deepmind/chex/blob/master/chex/_src/variants.py))
JAX relies extensively on code transformation and compilation, meaning that it
can be hard to ensure that code is properly tested. For instance, just testing a
python function using JAX code will not cover the actual code path that is
executed when jitted, and that path will also differ whether the code is jitted
for CPU, GPU, or TPU. This has been a source of obscure and hard to catch bugs
where XLA changes would lead to undesirable behaviours that however only
manifest in one specific code transformation.
Variants make it easy to ensure that unit tests cover different ‘variations’ of
a function, by providing a simple decorator that can be used to repeat any test
under all (or a subset) of the relevant code transformations.
E.g. suppose you want to test the output of a function `fn` with or without jit.
You can use `chex.variants` to run the test with both the jitted and non-jitted
version of the function by simply decorating a test method with
`@chex.variants`, and then using `self.variant(fn)` in place of `fn` in the body
of the test.
```python
def fn(x, y):
return x + y
...
class ExampleTest(chex.TestCase):
@chex.variants(with_jit=True, without_jit=True)
def test(self):
var_fn = self.variant(fn)
self.assertEqual(fn(1, 2), 3)
self.assertEqual(var_fn(1, 2), fn(1, 2))
```
If you define the function in the test method, you may also use `self.variant`
as a decorator in the function definition. For example:
```python
class ExampleTest(chex.TestCase):
@chex.variants(with_jit=True, without_jit=True)
def test(self):
@self.variant
def var_fn(x, y):
return x + y
self.assertEqual(var_fn(1, 2), 3)
```
Example of parameterized test:
```python
from absl.testing import parameterized
# Could also be:
# `class ExampleParameterizedTest(chex.TestCase, parameterized.TestCase):`
# `class ExampleParameterizedTest(chex.TestCase):`
class ExampleParameterizedTest(parameterized.TestCase):
@chex.variants(with_jit=True, without_jit=True)
@parameterized.named_parameters(
('case_positive', 1, 2, 3),
('case_negative', -1, -2, -3),
)
def test(self, arg_1, arg_2, expected):
@self.variant
def var_fn(x, y):
return x + y
self.assertEqual(var_fn(arg_1, arg_2), expected)
```
Chex currently supports the following variants:
* `with_jit` -- applies `jax.jit()` transformation to the function.
* `without_jit` -- uses the function as is, i.e. identity transformation.
* `with_device` -- places all arguments (except specified in `ignore_argnums`
argument) into device memory before applying the function.
* `without_device` -- places all arguments in RAM before applying the function.
* `with_pmap` -- applies `jax.pmap()` transformation to the function (see notes below).
See documentation in [variants.py](https://github.com/deepmind/chex/blob/master/chex/_src/variants.py) for more details on the supported variants.
More examples can be found in [variants_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/variants_test.py).
### Variants notes
* Test classes that use `@chex.variants` must inherit from
`chex.TestCase` (or any other base class that unrolls tests generators
within `TestCase`, e.g. `absl.testing.parameterized.TestCase`).
* **[`jax.vmap`]** All variants can be applied to a vmapped function;
please see an example in [variants_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/variants_test.py) (`test_vmapped_fn_named_params` and
`test_pmap_vmapped_fn`).
* **[`@chex.all_variants`]** You can get all supported variants
by using the decorator `@chex.all_variants`.
* **[`with_pmap` variant]** `jax.pmap(fn)`
([doc](https://jax.readthedocs.io/en/latest/jax.html#jax.pmap)) performs
parallel map of `fn` onto multiple devices. Since most tests run in a
single-device environment (i.e. having access to a single CPU or GPU), in which
case `jax.pmap` is a functional equivalent to `jax.jit`, ` with_pmap` variant is
skipped by default (although it works fine with a single device). Below we
describe a way to properly test `fn` if it is supposed to be used in
multi-device environments (TPUs or multiple CPUs/GPUs). To disable skipping
`with_pmap` variants in case of a single device, add
`--chex_skip_pmap_variant_if_single_device=false` to your test command.
### Fakes ([fake.py](https://github.com/deepmind/chex/blob/master/chex/_src/fake.py))
Debugging in JAX is made more difficult by code transformations such as `jit`
and `pmap`, which introduce optimizations that make code hard to inspect and
trace. It can also be difficult to disable those transformations during
debugging as they can be called at several places in the underlying
code. Chex provides tools to globally replace `jax.jit` with a no-op
transformation and `jax.pmap` with a (non-parallel) `jax.vmap`, in order to more
easily debug code in a single-device context.
For example, you can use Chex to fake `pmap` and have it replaced with a `vmap`.
This can be achieved by wrapping your code with a context manager:
```python
with chex.fake_pmap():
@jax.pmap
def fn(inputs):
...
# Function will be vmapped over inputs
fn(inputs)
```
The same functionality can also be invoked with `start` and `stop`:
```python
fake_pmap = chex.fake_pmap()
fake_pmap.start()
... your jax code ...
fake_pmap.stop()
```
In addition, you can fake a real multi-device test environment with a
multi-threaded CPU. See section **Faking multi-device test environments** for
more details.
See documentation in [fake.py](https://github.com/deepmind/chex/blob/master/chex/_src/fake.py) and examples in [fake_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/fake_test.py) for more details.
## Faking multi-device test environments
In situations where you do not have easy access to multiple devices, you can
still test parallel computation using single-device multi-threading.
In particular, one can force XLA to use a single CPU's threads as separate
devices, i.e. to fake a real multi-device environment with a multi-threaded one.
These two options are theoretically equivalent from XLA perspective because they
expose the same interface and use identical abstractions.
Chex has a flag `chex_n_cpu_devices` that specifies a number of CPU threads to
use as XLA devices.
To set up a multi-threaded XLA environment for `absl` tests, define
`setUpModule` function in your test module:
```python
def setUpModule():
chex.set_n_cpu_devices()
```
Now you can launch your test with `python test.py --chex_n_cpu_devices=N` to run
it in multi-device regime. Note that **all** tests within a module will have an
access to `N` devices.
More examples can be found in [variants_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/variants_test.py), [fake_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/fake_test.py) and [fake_set_n_cpu_devices_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/fake_set_n_cpu_devices_test.py).
### Using named dimension sizes.
Chex comes with a small utility that allows you to package a collection of
dimension sizes into a single object. The basic idea is:
```python
dims = chex.Dimensions(B=batch_size, T=sequence_len, E=embedding_dim)
...
chex.assert_shape(arr, dims['BTE'])
```
String lookups are translated integer tuples. For instance, let's say
`batch_size == 3`, `sequence_len = 5` and `embedding_dim = 7`, then
```python
dims['BTE'] == (3, 5, 7)
dims['B'] == (3,)
dims['TTBEE'] == (5, 5, 3, 7, 7)
...
```
You can also assign dimension sizes dynamically as follows:
```python
dims['XY'] = some_matrix.shape
dims.Z = 13
```
For more examples, see [chex.Dimensions](https://chex.readthedocs.io/en/latest/api.html#chex.Dimensions)
documentation.
## Citing Chex
This repository is part of the [DeepMind JAX Ecosystem], to cite Chex please use
the [DeepMind JAX Ecosystem citation].
[DeepMind JAX Ecosystem]: https://deepmind.com/blog/article/using-jax-to-accelerate-our-research "DeepMind JAX Ecosystem"
[DeepMind JAX Ecosystem citation]: https://github.com/deepmind/jax/blob/main/deepmind2020jax.txt "Citation"
Raw data
{
"_id": null,
"home_page": "https://github.com/deepmind/chex",
"name": "chex",
"maintainer": "",
"docs_url": null,
"requires_python": ">=3.9",
"maintainer_email": "",
"keywords": "jax testing debugging python machine learning",
"author": "DeepMind",
"author_email": "chex-dev@google.com",
"download_url": "https://files.pythonhosted.org/packages/26/a2/46649fb9f6a33cc7c2822161cc5481f0ffe5965fde1e6fc4c3003cd22323/chex-0.1.86.tar.gz",
"platform": null,
"description": "# Chex\n\n\n\n\n\nChex is a library of utilities for helping to write reliable JAX code.\n\nThis includes utils to help:\n\n* Instrument your code (e.g. assertions, warnings)\n* Debug (e.g. transforming `pmaps` in `vmaps` within a context manager).\n* Test JAX code across many `variants` (e.g. jitted vs non-jitted).\n\n## Installation\n\nYou can install the latest released version of Chex from PyPI via:\n\n```sh\npip install chex\n```\n\nor you can install the latest development version from GitHub:\n\n```sh\npip install git+https://github.com/deepmind/chex.git\n```\n\n## Modules Overview\n\n### Dataclass ([dataclass.py](https://github.com/deepmind/chex/blob/master/chex/_src/dataclass.py))\n\nDataclasses are a popular construct introduced by Python 3.7 to allow to\neasily specify typed data structures with minimal boilerplate code. They are\nnot, however, compatible with JAX and\n[dm-tree](https://github.com/deepmind/tree) out of the box.\n\nIn Chex we provide a JAX-friendly dataclass implementation reusing python [dataclasses](https://docs.python.org/3/library/dataclasses.html#module-dataclasses).\n\nChex implementation of `dataclass` registers dataclasses as internal [_PyTree_\nnodes](https://jax.readthedocs.io/en/latest/pytrees.html) to ensure\ncompatibility with JAX data structures.\n\nIn addition, we provide a class wrapper that exposes dataclasses as\n`collections.Mapping` descendants which allows to process them\n(e.g. (un-)flatten) in `dm-tree` methods as usual Python dictionaries.\nSee [`@mappable_dataclass`](https://github.com/deepmind/chex/blob/master/chex/_src/dataclass.py#L27)\ndocstring for more details.\n\nExample:\n\n```python\n@chex.dataclass\nclass Parameters:\n x: chex.ArrayDevice\n y: chex.ArrayDevice\n\nparameters = Parameters(\n x=jnp.ones((2, 2)),\n y=jnp.ones((1, 2)),\n)\n\n# Dataclasses can be treated as JAX pytrees\njax.tree_util.tree_map(lambda x: 2.0 * x, parameters)\n\n# and as mappings by dm-tree\ntree.flatten(parameters)\n```\n\n**NOTE**: Unlike standard Python 3.7 dataclasses, Chex\ndataclasses cannot be constructed using positional arguments. They support\nconstruction arguments provided in the same format as the Python dict\nconstructor. Dataclasses can be converted to tuples with the `from_tuple` and\n`to_tuple` methods if necessary.\n\n```python\nparameters = Parameters(\n jnp.ones((2, 2)),\n jnp.ones((1, 2)),\n)\n# ValueError: Mappable dataclass constructor doesn't support positional args.\n```\n\n### Assertions ([asserts.py](https://github.com/deepmind/chex/blob/master/chex/_src/asserts.py))\n\nOne limitation of PyType annotations for JAX is that they do not support the\nspecification of `DeviceArray` ranks, shapes or dtypes. Chex includes a number\nof functions that allow flexible and concise specification of these properties.\n\nE.g. suppose you want to ensure that all tensors `t1`, `t2`, `t3` have the same\nshape, and that tensors `t4`, `t5` have rank `2` and (`3` or `4`), respectively.\n\n```python\nchex.assert_equal_shape([t1, t2, t3])\nchex.assert_rank([t4, t5], [2, {3, 4}])\n```\n\nMore examples:\n\n```python\nfrom chex import assert_shape, assert_rank, ...\n\nassert_shape(x, (2, 3)) # x has shape (2, 3)\nassert_shape([x, y], [(), (2,3)]) # x is scalar and y has shape (2, 3)\n\nassert_rank(x, 0) # x is scalar\nassert_rank([x, y], [0, 2]) # x is scalar and y is a rank-2 array\nassert_rank([x, y], {0, 2}) # x and y are scalar OR rank-2 arrays\n\nassert_type(x, int) # x has type `int` (x can be an array)\nassert_type([x, y], [int, float]) # x has type `int` and y has type `float`\n\nassert_equal_shape([x, y, z]) # x, y, and z have equal shapes\n\nassert_trees_all_close(tree_x, tree_y) # values and structure of trees match\nassert_tree_all_finite(tree_x) # all tree_x leaves are finite\n\nassert_devices_available(2, 'gpu') # 2 GPUs available\nassert_tpu_available() # at least 1 TPU available\n\nassert_numerical_grads(f, (x, y), j) # f^{(j)}(x, y) matches numerical grads\n```\n\nSee `asserts.py`\n[documentation](https://chex.readthedocs.io/en/latest/api.html#assertions) to\nfind all supported assertions.\n\nIf you cannot find a specific assertion, please consider making a pull request\nor openning an issue on\n[the bug tracker](https://github.com/deepmind/chex/issues).\n\n#### Optional Arguments\n\nAll chex assertions support the following optional kwargs for manipulating the\nemitted exception messages:\n\n* `custom_message`: A string to include into the emitted exception messages.\n* `include_default_message`: Whether to include the default Chex message into\n the emitted exception messages.\n* `exception_type`: An exception type to use. `AssertionError` by default.\n\nFor example, the following code:\n\n```python\ndataset = load_dataset()\nparams = init_params()\nfor i in range(num_steps):\n params = update_params(params, dataset.sample())\n chex.assert_tree_all_finite(params,\n custom_message=f'Failed at iteration {i}.',\n exception_type=ValueError)\n```\n\nwill raise a `ValueError` that includes a step number when `params` get polluted\nwith `NaNs` or `None`s.\n\n#### Static and Value (aka *Runtime*) Assertions\n\nChex divides all assertions into 2 classes: ***static*** and ***value***\nassertions.\n\n1. ***static*** assertions use anything except concrete values of tensors.\n Examples: `assert_shape`, `assert_trees_all_equal_dtypes`,\n `assert_max_traces`.\n\n2. ***value*** assertions require access to tensor values, which are not\n available during JAX tracing (see\n [HowJAX primitives work](https://jax.readthedocs.io/en/latest/notebooks/How_JAX_primitives_work.html)),\n thus such assertion need special treatment in a *jitted* code.\n\nTo enable value assertions in a jitted function, it can be decorated with\n`chex.chexify()` wrapper. Example:\n\n```python\n @chex.chexify\n @jax.jit\n def logp1_abs_safe(x: chex.Array) -> chex.Array:\n chex.assert_tree_all_finite(x)\n return jnp.log(jnp.abs(x) + 1)\n\n logp1_abs_safe(jnp.ones(2)) # OK\n logp1_abs_safe(jnp.array([jnp.nan, 3])) # FAILS (in async mode)\n\n # The error will be raised either at the next line OR at the next\n # `logp1_abs_safe` call. See the docs for more detain on async mode.\n logp1_abs_safe.wait_checks() # Wait for the (async) computation to complete.\n```\n\nSee\n[this docstring](https://chex.readthedocs.io/en/latest/api.html#chex.chexify)\nfor more detail on `chex.chexify()`.\n\n#### JAX Tracing Assertions\n\nJAX re-traces JIT'ted function every time the structure of passed arguments\nchanges. Often this behavior is inadvertent and leads to a significant\nperformance drop which is hard to debug. [@chex.assert_max_traces](https://github.com/deepmind/chex/blob/master/chex/_src/asserts.py#L44)\ndecorator asserts that the function is not re-traced more than `n` times during\nprogram execution.\n\nGlobal trace counter can be cleared by calling\n`chex.clear_trace_counter()`. This function be used to isolate unittests relying\non `@chex.assert_max_traces`.\n\nExamples:\n\n```python\n @jax.jit\n @chex.assert_max_traces(n=1)\n def fn_sum_jitted(x, y):\n return x + y\n\n fn_sum_jitted(jnp.zeros(3), jnp.zeros(3)) # tracing for the 1st time - OK\n fn_sum_jitted(jnp.zeros([6, 7]), jnp.zeros([6, 7])) # AssertionError!\n```\n\nCan be used with `jax.pmap()` as well:\n\n```python\n def fn_sub(x, y):\n return x - y\n\n fn_sub_pmapped = jax.pmap(chex.assert_max_traces(fn_sub, n=10))\n```\n\nSee\n[HowJAX primitives work](https://jax.readthedocs.io/en/latest/notebooks/How_JAX_primitives_work.html)\nsection for more information about tracing.\n\n### Warnings ([warnigns.py](https://github.com/deepmind/chex/blob/master/chex/_src/warnings.py))\n\nIn addition to hard assertions Chex also offers utilities to add common\nwarnings, such as specific types of deprecation warnings.\n\n### Test variants ([variants.py](https://github.com/deepmind/chex/blob/master/chex/_src/variants.py))\n\nJAX relies extensively on code transformation and compilation, meaning that it\ncan be hard to ensure that code is properly tested. For instance, just testing a\npython function using JAX code will not cover the actual code path that is\nexecuted when jitted, and that path will also differ whether the code is jitted\nfor CPU, GPU, or TPU. This has been a source of obscure and hard to catch bugs\nwhere XLA changes would lead to undesirable behaviours that however only\nmanifest in one specific code transformation.\n\nVariants make it easy to ensure that unit tests cover different \u2018variations\u2019 of\na function, by providing a simple decorator that can be used to repeat any test\nunder all (or a subset) of the relevant code transformations.\n\nE.g. suppose you want to test the output of a function `fn` with or without jit.\nYou can use `chex.variants` to run the test with both the jitted and non-jitted\nversion of the function by simply decorating a test method with\n`@chex.variants`, and then using `self.variant(fn)` in place of `fn` in the body\nof the test.\n\n```python\ndef fn(x, y):\n return x + y\n...\n\nclass ExampleTest(chex.TestCase):\n\n @chex.variants(with_jit=True, without_jit=True)\n def test(self):\n var_fn = self.variant(fn)\n self.assertEqual(fn(1, 2), 3)\n self.assertEqual(var_fn(1, 2), fn(1, 2))\n```\n\nIf you define the function in the test method, you may also use `self.variant`\nas a decorator in the function definition. For example:\n\n```python\nclass ExampleTest(chex.TestCase):\n\n @chex.variants(with_jit=True, without_jit=True)\n def test(self):\n @self.variant\n def var_fn(x, y):\n return x + y\n\n self.assertEqual(var_fn(1, 2), 3)\n```\n\nExample of parameterized test:\n\n```python\nfrom absl.testing import parameterized\n\n# Could also be:\n# `class ExampleParameterizedTest(chex.TestCase, parameterized.TestCase):`\n# `class ExampleParameterizedTest(chex.TestCase):`\nclass ExampleParameterizedTest(parameterized.TestCase):\n\n @chex.variants(with_jit=True, without_jit=True)\n @parameterized.named_parameters(\n ('case_positive', 1, 2, 3),\n ('case_negative', -1, -2, -3),\n )\n def test(self, arg_1, arg_2, expected):\n @self.variant\n def var_fn(x, y):\n return x + y\n\n self.assertEqual(var_fn(arg_1, arg_2), expected)\n```\n\nChex currently supports the following variants:\n\n* `with_jit` -- applies `jax.jit()` transformation to the function.\n* `without_jit` -- uses the function as is, i.e. identity transformation.\n* `with_device` -- places all arguments (except specified in `ignore_argnums`\n argument) into device memory before applying the function.\n* `without_device` -- places all arguments in RAM before applying the function.\n* `with_pmap` -- applies `jax.pmap()` transformation to the function (see notes below).\n\nSee documentation in [variants.py](https://github.com/deepmind/chex/blob/master/chex/_src/variants.py) for more details on the supported variants.\nMore examples can be found in [variants_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/variants_test.py).\n\n### Variants notes\n\n* Test classes that use `@chex.variants` must inherit from\n`chex.TestCase` (or any other base class that unrolls tests generators\nwithin `TestCase`, e.g. `absl.testing.parameterized.TestCase`).\n\n* **[`jax.vmap`]** All variants can be applied to a vmapped function;\nplease see an example in [variants_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/variants_test.py) (`test_vmapped_fn_named_params` and\n`test_pmap_vmapped_fn`).\n\n* **[`@chex.all_variants`]** You can get all supported variants\nby using the decorator `@chex.all_variants`.\n\n* **[`with_pmap` variant]** `jax.pmap(fn)`\n([doc](https://jax.readthedocs.io/en/latest/jax.html#jax.pmap)) performs\nparallel map of `fn` onto multiple devices. Since most tests run in a\nsingle-device environment (i.e. having access to a single CPU or GPU), in which\ncase `jax.pmap` is a functional equivalent to `jax.jit`, ` with_pmap` variant is\nskipped by default (although it works fine with a single device). Below we\ndescribe a way to properly test `fn` if it is supposed to be used in\nmulti-device environments (TPUs or multiple CPUs/GPUs). To disable skipping\n`with_pmap` variants in case of a single device, add\n`--chex_skip_pmap_variant_if_single_device=false` to your test command.\n\n### Fakes ([fake.py](https://github.com/deepmind/chex/blob/master/chex/_src/fake.py))\n\nDebugging in JAX is made more difficult by code transformations such as `jit`\nand `pmap`, which introduce optimizations that make code hard to inspect and\ntrace. It can also be difficult to disable those transformations during\ndebugging as they can be called at several places in the underlying\ncode. Chex provides tools to globally replace `jax.jit` with a no-op\ntransformation and `jax.pmap` with a (non-parallel) `jax.vmap`, in order to more\neasily debug code in a single-device context.\n\nFor example, you can use Chex to fake `pmap` and have it replaced with a `vmap`.\nThis can be achieved by wrapping your code with a context manager:\n\n```python\nwith chex.fake_pmap():\n @jax.pmap\n def fn(inputs):\n ...\n\n # Function will be vmapped over inputs\n fn(inputs)\n```\n\nThe same functionality can also be invoked with `start` and `stop`:\n\n```python\nfake_pmap = chex.fake_pmap()\nfake_pmap.start()\n... your jax code ...\nfake_pmap.stop()\n```\n\nIn addition, you can fake a real multi-device test environment with a\nmulti-threaded CPU. See section **Faking multi-device test environments** for\nmore details.\n\nSee documentation in [fake.py](https://github.com/deepmind/chex/blob/master/chex/_src/fake.py) and examples in [fake_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/fake_test.py) for more details.\n\n## Faking multi-device test environments\n\nIn situations where you do not have easy access to multiple devices, you can\nstill test parallel computation using single-device multi-threading.\n\nIn particular, one can force XLA to use a single CPU's threads as separate\ndevices, i.e. to fake a real multi-device environment with a multi-threaded one.\nThese two options are theoretically equivalent from XLA perspective because they\nexpose the same interface and use identical abstractions.\n\nChex has a flag `chex_n_cpu_devices` that specifies a number of CPU threads to\nuse as XLA devices.\n\nTo set up a multi-threaded XLA environment for `absl` tests, define\n`setUpModule` function in your test module:\n\n```python\ndef setUpModule():\n chex.set_n_cpu_devices()\n```\n\nNow you can launch your test with `python test.py --chex_n_cpu_devices=N` to run\nit in multi-device regime. Note that **all** tests within a module will have an\naccess to `N` devices.\n\nMore examples can be found in [variants_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/variants_test.py), [fake_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/fake_test.py) and [fake_set_n_cpu_devices_test.py](https://github.com/deepmind/chex/blob/master/chex/_src/fake_set_n_cpu_devices_test.py).\n\n### Using named dimension sizes.\n\nChex comes with a small utility that allows you to package a collection of\ndimension sizes into a single object. The basic idea is:\n\n```python\ndims = chex.Dimensions(B=batch_size, T=sequence_len, E=embedding_dim)\n...\nchex.assert_shape(arr, dims['BTE'])\n```\n\nString lookups are translated integer tuples. For instance, let's say\n`batch_size == 3`, `sequence_len = 5` and `embedding_dim = 7`, then\n\n```python\ndims['BTE'] == (3, 5, 7)\ndims['B'] == (3,)\ndims['TTBEE'] == (5, 5, 3, 7, 7)\n...\n```\n\nYou can also assign dimension sizes dynamically as follows:\n\n```python\ndims['XY'] = some_matrix.shape\ndims.Z = 13\n```\n\nFor more examples, see [chex.Dimensions](https://chex.readthedocs.io/en/latest/api.html#chex.Dimensions)\ndocumentation.\n\n## Citing Chex\n\nThis repository is part of the [DeepMind JAX Ecosystem], to cite Chex please use\nthe [DeepMind JAX Ecosystem citation].\n\n[DeepMind JAX Ecosystem]: https://deepmind.com/blog/article/using-jax-to-accelerate-our-research \"DeepMind JAX Ecosystem\"\n[DeepMind JAX Ecosystem citation]: https://github.com/deepmind/jax/blob/main/deepmind2020jax.txt \"Citation\"\n",
"bugtrack_url": null,
"license": "Apache 2.0",
"summary": "Chex: Testing made fun, in JAX!",
"version": "0.1.86",
"project_urls": {
"Homepage": "https://github.com/deepmind/chex"
},
"split_keywords": [
"jax",
"testing",
"debugging",
"python",
"machine",
"learning"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "e6ed625d545d08c6e258d2d63a93a0bf8ed8a296c09d254208e73f9d4fb0b746",
"md5": "98325680002cab223d742265057160e5",
"sha256": "251c20821092323a3d9c28e1cf80e4a58180978bec368f531949bd9847eee568"
},
"downloads": -1,
"filename": "chex-0.1.86-py3-none-any.whl",
"has_sig": false,
"md5_digest": "98325680002cab223d742265057160e5",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.9",
"size": 98167,
"upload_time": "2024-03-19T10:05:26",
"upload_time_iso_8601": "2024-03-19T10:05:26.685004Z",
"url": "https://files.pythonhosted.org/packages/e6/ed/625d545d08c6e258d2d63a93a0bf8ed8a296c09d254208e73f9d4fb0b746/chex-0.1.86-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "26a246649fb9f6a33cc7c2822161cc5481f0ffe5965fde1e6fc4c3003cd22323",
"md5": "c4255e641b7bdb549038174ab4e74ce8",
"sha256": "e8b0f96330eba4144659e1617c0f7a57b161e8cbb021e55c6d5056c7378091d1"
},
"downloads": -1,
"filename": "chex-0.1.86.tar.gz",
"has_sig": false,
"md5_digest": "c4255e641b7bdb549038174ab4e74ce8",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.9",
"size": 89021,
"upload_time": "2024-03-19T10:05:29",
"upload_time_iso_8601": "2024-03-19T10:05:29.343498Z",
"url": "https://files.pythonhosted.org/packages/26/a2/46649fb9f6a33cc7c2822161cc5481f0ffe5965fde1e6fc4c3003cd22323/chex-0.1.86.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-03-19 10:05:29",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "deepmind",
"github_project": "chex",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "chex"
}
JSON
