Home » Oreps » Jax » All Releases » jax-v0.2.21 Release
google/jax
Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
Watch
1
Commit Activity
Directory Browser
Velocity History
Releasesjax-v0.2.21
Basic Details Issues Addressed Top Contributors Directory Browser Release Notes
Jax jax-v0.2.21

Jax: jax-v0.2.21 Release

Release date:
September 23, 2021
Previous version:
jax-v0.2.20 (released September 3, 2021)
Magnitude:
10,380 Diff Delta
Contributors:
24 total committers
Data confidence:
Commits:
110 impacting commits

110 Commits in this Release

Ordered by the degree to which they evolved the repo in this version.

Refactor JAX CPU kernels to make them usable from C++.
Authored September 3, 2021
Use jax.* APIs rather than api.* names in tests.
Authored September 13, 2021
Add more @jit decorators to jax.numpy functions.
Authored September 9, 2021
Move contents of jax.lib to jax._src.lib.
Authored September 23, 2021
Combine initial_style_batchers with collective_rules
Authored August 26, 2021
Turn jnp.ndarray into a true abstract base class.
Authored September 15, 2021
Switch internal users of jax.util.partial to use functools.partial.
Authored September 13, 2021
Use shorter variable names in lax.py.
Authored September 8, 2021
[sparse] add support for sparse-sparse matrix products
Authored September 3, 2021
New lowering APIs for pjit
Authored September 8, 2021
Attempt to land https://github.com/google/jax/pull/6400 again.
Authored September 21, 2021
Don't cast out_axis_resources to a tuple automatically
Authored September 7, 2021
Add PEP484-compatible export for jax and its subpackages
Authored August 30, 2021
Add a @jit decorator around jnp.linspace.
Authored September 7, 2021
Fix package exports
Authored September 14, 2021
[jax2tf] Fix stale comments
Authored September 12, 2021
[sparse] add BCOO._dedupe() method
Authored September 3, 2021
[host_callback] Fix an assertion failure for grad(remat(host_callback))
Authored September 17, 2021
jax.numpy: add wrapper for np.insert
Authored September 16, 2021
Move ShardedDeviceArray & PmapFunction to the raw C API and implement pickling/unpickling.
Authored September 7, 2021
Feat: scipy.stats.nbinom implementation
Authored September 8, 2021
Remove axis name from named_shape when unmapping avals
Authored September 8, 2021
Add new lowering APIs for jit
Authored September 9, 2021
Implement variance scaling initializers with complex dtype
Authored July 31, 2021
Move contents of jax.api_util to jax._src.api_util and add a forwarding shim.
Authored September 8, 2021
[jax2tf] Ensure that shared constants in JAX are shared also in the converted function
Authored September 23, 2021
[sparse] add BCOO._unbatch() utility
Authored September 3, 2021
Ensure that named axes consistently refer to global axis sizes in xmap
Authored September 7, 2021
Remove jax.api.
Authored September 16, 2021
mark custom derivative calls as API boundaries for stack trace filtering
Authored September 21, 2021
convolutions: use flip() to clean up reverse-indexing
Authored September 21, 2021
jnp.array: raise TypeError on boolean scalar indices
Authored September 15, 2021
handle dropped output values in the sparse interpreter
Authored September 8, 2021
Delete Cuda 10.2 release and continuous builds.
Authored September 3, 2021
bazelrc changes for building Cuda 11.4 wheels with cudnn 8.2
Authored September 16, 2021
Fix wrong dtype output from indexing with an empty slice.
Authored September 22, 2021
handle linear custom_jvp functions
Authored September 3, 2021
Expose allow_spmd_sharding_propagation_to_output in Python
Authored September 13, 2021
Replace uses of jax.partial with functools.partial, in preparation for removing jax.partial.
Authored September 13, 2021
Cache used_axis_names calls for jaxprs
Authored September 10, 2021

Browse Other Releases

Latest Pending Unreleased 😎
jax-v0.2.26 Released December 8, 2021
1,528 Ξ”
jax-v0.2.25 Released October 25, 2021
2,029 Ξ”
jax-v0.2.24 Released October 19, 2021
298 Ξ”
jax-v0.2.23 Released October 19, 2021
0 Ξ”
jax-v0.2.22 Released October 13, 2021
37 Ξ”
jax-v0.2.21 Released September 23, 2021
10,380 Ξ”
jax-v0.2.20 Released September 3, 2021
0 Ξ”
jax-v0.2.19 Released August 13, 2021
12,452 Ξ”
jax-v0.2.18 Released July 21, 2021
0 Ξ”
jax-v0.2.17 Released July 9, 2021
6,049 Ξ”

Top Contributors in jax-v0.2.21

hawkinsp
jakevdp
apaszke
sharadmv
gnecula
froystig
yashk2810
jblespiau
cyprienc
wdphy16

Directory Browser for jax-v0.2.21

We haven't yet finished calculating and confirming the files and directories changed in this release. Please check back soon.

Release Notes Published

  • New features:

    • Added jax.numpy.insert implementation (#7936 ).

  • Breaking Changes

    • jax.api has been removed. Functions that were available as jax.api.* were aliases for functions in jax.*; please use the functions in jax.* instead.
    • jax.partial, jax.lax.partial, and jax.util.partial were accidental exports that have now been removed. Use functools.partial from the Python standard library instead.
    • Boolean scalar indices now raise a TypeError; previously this silently returned wrong results (#7925 ).
    • Many more jax.numpy functions now require array-like inputs, and will error if passed a list (#7747 #7802 #7907 ). See #7737 for a discussion of the rationale behind this change.
    • When inside a transformation such as jax.jit, jax.numpy.array always stages the array it produces into the traced computation. Previously jax.numpy.array would sometimes produce a on-device array, even under a jax.jit decorator. This change may break code that used JAX arrays to perform shape or index computations that must be known statically; the workaround is to perform such computations using classic NumPy arrays instead.
    • jnp.ndarray is now a true base-class for JAX arrays. In particular, this means that for a standard numpy array x, isinstance(x, jnp.ndarray) will now return False (#7927).
GitClear uses cookies to ensure you get the best experience on our website. Learn more