Home » Oreps » Jax » All Releases » jax-v0.2.26 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-v0.2.26
Basic Details Issues Addressed Top Contributors Directory Browser Release Notes
Jax jax-v0.2.26

Jax: jax-v0.2.26 Release

Release date:
December 8, 2021
Previous version:
jax-v0.2.25 (released October 25, 2021)
Magnitude:
1,528 Diff Delta
Contributors:
6 total committers
Data confidence:
Commits:
24 impacting commits

24 Commits in this Release

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

[sparse] add todense() primitive for use in sparsify transform
Authored October 26, 2021
make iter(DeviceArray) return DeviceArrays w/o sync
Authored July 22, 2020
[sparse] implement broadcasted sparse-dense multiplication
Authored October 21, 2021
[sparse] add sparse.random_bcoo() utility
Authored October 26, 2021
Add jax.distributed.initialize for multi-host GPU.
Authored October 25, 2021
[sparse] avoid default jit-compilation of sparse array functions
Authored October 26, 2021
jit(f).lower(...) works w/ duck typed shape/dtype
Authored October 28, 2021
benchmarks: add compilation of sparse operations
Authored October 26, 2021
lax_numpy: replace some reshapes with expand_dims
Authored October 28, 2021
benchmarks: add JIT versions of sparse.BCOO benchmarks
Authored October 26, 2021
[XLA:TPU] Fix shape bug for k=1 in ApproxTopK
Authored October 26, 2021
Use `SourceInfo` named tuple to keep track of source information
Authored October 27, 2021
benchmarks: add sparse.BCOO todense/fromdense/matvec benchmarks
Authored October 25, 2021
Broadcast unmapped values in all_to_all batching rule
Authored October 6, 2021
Support multiple in/out axes in scaled inits
Authored October 20, 2021
add functionality test
Authored October 20, 2021
readthedocs: use new build configuration & update to Python 3.9
Authored October 28, 2021
MeshComputation.hlo should be a method
Authored October 27, 2021
[JAX] Update JAX users in preparation for a change that makes iteration over a JAX array return JAX arrays, instead of NumPy arrays.
Authored October 27, 2021
dynamic_slice: ensure start_indices dtypes match
Authored October 20, 2021
[JAX] Polish doc formatting for approx_top_k.
Authored October 26, 2021
fix typo
Authored October 28, 2021
update docstring
Authored October 20, 2021
DOC: pin jupyter-core to fix RTD build
Authored October 27, 2021

Browse Other Releases

Latest Pending Unreleased 😎
jax-v0.3.1 Released February 18, 2022
6,721 Ξ”
jax-v0.3.0 Released February 10, 2022
5,978 Ξ”
jax-v0.2.28 Released February 2, 2022
829 Ξ”
jax-v0.2.27 Released January 18, 2022
6,502 Ξ”
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”

Top Contributors in jax-v0.2.26

jakevdp
mattjj
zhangqiaorjc
sharadmv
a-googler
jekbradbury

Directory Browser for jax-v0.2.26

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

Release Notes Published

  • Bug fixes:

  • Out-of-bounds indices to jax.ops.segment_sum will now be handled with FILL_OR_DROP semantics, as documented. This primarily afects the reverse-mode derivative, where gradients corresponding to out-of-bounds indices will now be returned as 0. (#8634).

  • jax2tf will force the converted code to use XLA for the code fragments under jax.jit, e.g., most jax.numpy functions (#7839).

GitClear uses cookies to ensure you get the best experience on our website. Learn more