Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
611 commits
Select commit Hold shift + click to select a range
aef4637
Update XLA dependency to use revision http://github.com/openxla/xla/c…
Google-ML-Automation Oct 21, 2025
0e61c8e
Minor typo fix in _non_static_arg_names docstring
vfdev-5 Oct 21, 2025
a07a612
[Pallas] Explicitly use Triton in export tests wherever the use was i…
bchetioui Oct 21, 2025
fe2316e
[Mosaic GPU] Cleanup and fix a few things for the `Divides` constraint.
dimitar-asenov Oct 21, 2025
7489c79
[Pallas:MGPU] Substantially refactor the PyTorch conversion logic
apaszke Oct 21, 2025
71be58f
Correct typo in docstring for with_sharding_constraint
vfdev-5 Oct 21, 2025
104c747
Merge pull request #32772 from vfdev-5:typo-fix-_non_static_arg_names
Google-ML-Automation Oct 21, 2025
6a0d2c2
[mosaic] Minor improvements to the tpu.assume_multiple op
superbobry Oct 21, 2025
c6827d5
Document wrap argument of fill_diagonal.
carlosgmartin Oct 21, 2025
17ca886
Remove ffi::RemainingArgs from CPU triangular solve kernel.
hawkinsp Oct 21, 2025
459cb05
Downgrade wheel in requirement lock files as wheel==0.46.1 was yanked
kanglant Oct 21, 2025
67fef15
Merge pull request #32770 from carlosgmartin:fill_diagonal_wrap_docum…
Google-ML-Automation Oct 21, 2025
1e55aff
Remove jax2tf multiprocess GPU/TPU tests from CI (recently added erro…
emilyfertig Oct 21, 2025
e9a85f2
Merge pull request #32783 from kanglant:update_req
Google-ML-Automation Oct 21, 2025
a43badd
Permit a different error message in a multiprocess array test.
emilyfertig Oct 21, 2025
178e622
Fixed residual issue for jax.nn.dot_product_attention
Google-ML-Automation Oct 21, 2025
1dd7899
Merge pull request #31433 from jaro-sevcik:scaled-matmul-handle-overl…
Google-ML-Automation Oct 21, 2025
36dd04e
[Mosaic] Increase the number of partial accumulators used in multi_re…
yueshengys Oct 21, 2025
d5eb954
Improve the documentation of dot_product_attention. Add bias matrix t…
carlosgmartin Oct 21, 2025
8e5fc05
[mosaic] lower pltpu.repeat to jnp.tile
Google-ML-Automation Oct 21, 2025
b0664df
Remove redundant cast in `_dot_general_transpose_rhs`.
chr1sj0nes Oct 22, 2025
7963bcc
Merge pull request #32742 from gnecula:err_msg_ser
Google-ML-Automation Oct 22, 2025
d5c78b3
Check all environment variables for Slurm environment
Flamefire Oct 22, 2025
fae227a
Update XLA dependency to use revision http://github.com/openxla/xla/c…
Google-ML-Automation Oct 22, 2025
94c7147
Removed a few unused BUILD deps
superbobry Oct 22, 2025
255c517
Fix the interaction of `jit(vmap(grad(shmap)))` under sharding-in-types.
yashk2810 Oct 22, 2025
65979a9
Disable sparse_bcoo_bcsr_test on TPU 7x since it times out
yashk2810 Oct 22, 2025
5466dd7
Relax shape check in `_compute_fans`.
Google-ML-Automation Oct 22, 2025
22439d6
[doc] improve docs on inputs to jnp.array/asarray
jakevdp Oct 22, 2025
907c7c2
[Mosaic TPU] Support stochastic rounding on older generations.
megrez-yliu Oct 22, 2025
ac996df
[pallas:mgpu] Fixed a failure in `PallasCallRemoteDMATest`
superbobry Oct 22, 2025
e828579
Reverts 96ce5992bb9ca33265691268f7e98a3762c523c5
hawkinsp Oct 22, 2025
a47e4cc
Merge pull request #32808 from jakevdp:fix-array-doc
Google-ML-Automation Oct 22, 2025
a35ce56
Strip one-sized mesh axis from vma bits if `remove_size_one_mesh_axis…
yashk2810 Oct 22, 2025
3ad942d
Remove usage of mirrored `tar` files from CI because hermetic `xz` to…
ybaturina Oct 22, 2025
737eb82
Deactivate GPU multiprocesses tests in the freethreading build since …
emilyfertig Oct 22, 2025
edbfcaa
Update/skip Pallas tests for tpu v7x
justinjfu Oct 22, 2025
0fd3b05
Handle numpy/jax in_axis argument in _compute_fans.
Google-ML-Automation Oct 22, 2025
86e5878
Deactivate multiprocess array_test_cpu since it fails on windows, arm…
emilyfertig Oct 22, 2025
2c5a5a2
Fix the manual profile capture option documentation.
Google-ML-Automation Oct 22, 2025
dc18664
Disable tsan for multiprocess tests with flaky timeouts.
emilyfertig Oct 22, 2025
e1d1cbf
[jax:ffi] Do not pass uninitialized type_id value to type registratio…
ezhulenev Oct 22, 2025
f858e96
Add new atomic `live_devices` API.
mwhittaker Oct 23, 2025
2f8207d
Fix `pl.debug_print` passing extraneous scalar placeholders to TPU lo…
Google-ML-Automation Oct 23, 2025
38dfac1
[Mosaic GPU] Print operation on load/store validation failures.
allanrenucci Oct 23, 2025
5902264
[Mosaic GPU][NFC] Simplify `test_wgmma_kernel_with_tma` with `paramet…
allanrenucci Oct 23, 2025
cab8c22
Allow slice_sizes to gather that would exceed the operand dimension i…
hawkinsp Oct 23, 2025
c890ac5
Improve the error message that is raised in pvary and psum_invariant …
yashk2810 Oct 23, 2025
3cc2907
Disable tsan for array_extensibility_test since it times out
yashk2810 Oct 23, 2025
671483f
Always load MPMD dialect during serialization/de-serialization
alhaad Oct 23, 2025
e4b68b2
Update XLA dependency to use revision http://github.com/openxla/xla/c…
Google-ML-Automation Oct 23, 2025
ad62c67
Merge pull request #32777 from vfdev-5:patch-4
Google-ML-Automation Oct 23, 2025
f26e8c6
[pallas:sc] Fixed a typo in `_store_lowering_rule`
superbobry Oct 23, 2025
7783bef
Merge pull request #32799 from Flamefire:patch-1
Google-ML-Automation Oct 23, 2025
3fd4597
Merge pull request #30485 from pfackeldey:patch-1
Google-ML-Automation Oct 23, 2025
1c51987
Merge pull request #32715 from nurmukhametov:fix-square-todo
Google-ML-Automation Oct 23, 2025
835e225
Merge pull request #32046 from tripathiji1312:cuda_version_descripency
Google-ML-Automation Oct 23, 2025
1ae6899
Merge pull request #28549 from jjyyxx:patch-4
Google-ML-Automation Oct 23, 2025
37cd873
Merge pull request #32789 from carlosgmartin:dot_product_attention_fo…
Google-ML-Automation Oct 23, 2025
5d5f744
Merge pull request #27292 from uuirs:fix_decode_attention
Google-ML-Automation Oct 23, 2025
38637b3
Merge pull request #26682 from alhridoy:docs-debug-print-perturbation
Google-ML-Automation Oct 23, 2025
3be154c
Merge pull request #24136 from jjyyxx:patch-3
Google-ML-Automation Oct 23, 2025
bfed503
Merge pull request #24350 from jakevdp:eig-grad-error
Google-ML-Automation Oct 23, 2025
49d4367
Fixed the jupytext failure at head
superbobry Oct 23, 2025
fe76261
Merge pull request #32830 from superbobry:main
Google-ML-Automation Oct 23, 2025
fe82287
[pallas:mosaic] Added type annotations for a few primitives
superbobry Oct 23, 2025
870aa9b
Reverts 36dd04ebd317133b1127faa2ff211291494e3ee7
Google-ML-Automation Oct 23, 2025
72c445a
Disable ASAN for a gpu test that's failing due to a memory leak in NCCL.
emilyfertig Oct 23, 2025
50f48ed
[docs] Manually imported https://github.com/jax-ml/jax/pull/20549
superbobry Oct 23, 2025
87bccb7
Manually imported https://github.com/jax-ml/jax/pull/20791
superbobry Oct 23, 2025
ab8e296
[doc] update python/numpy/scipy version support table
jakevdp Oct 23, 2025
3dd28ce
[pmap] Make sure `cpp_pmap` gets tested as well, if only internally, …
danielsuo Oct 23, 2025
702bd8c
Tighten numpy.linalg.qr annotation
NeilGirdhar May 24, 2025
1160b89
Merge pull request #32844 from jakevdp:version-support
Google-ML-Automation Oct 23, 2025
0dffef2
Fixup Potential solution 2 Custom pytree init
ravwojdyla Jun 10, 2025
1467437
Merge pull request #28997 from NeilGirdhar:annotation
Google-ML-Automation Oct 23, 2025
35cde24
[pallas:sc] Added a test for lax.split
superbobry Oct 23, 2025
b93bb7d
Added shutdown timeout to `jax.distributed.initialize`.
mwhittaker Oct 23, 2025
8d58e7c
Ensure that the weakref cache's LRU list outlives the LRU caches that…
hawkinsp Oct 23, 2025
4e5a9af
Error out if unreduced or reduced is an input to any collective.
yashk2810 Oct 23, 2025
9a61e09
[JAX] Add compiled program call_location logging support to JAX
Google-ML-Automation Oct 23, 2025
7a5c5a1
[MOSAIC] Use shape invariant numerics by default and opt into shape v…
blakehechtman Oct 23, 2025
bc5fc36
Merge pull request #29360 from ravwojdyla:rav-fixup-doc
Google-ML-Automation Oct 23, 2025
8c5f4a3
[mutable-arrays] fix vjp3 fancy transpose bug
mattjj Oct 23, 2025
ea61015
Update XLA dependency to use revision http://github.com/openxla/xla/c…
Google-ML-Automation Oct 24, 2025
15e9723
Merge pull request #32853 from mattjj:fix-pjit-transpose-bug
Google-ML-Automation Oct 24, 2025
b815692
[pallas:sc] Error on direct loads/stores from/to HBM and VMEM_SHARED
superbobry Oct 24, 2025
9366503
Integrate LLVM at llvm/llvm-project@917d1f20aecf
Google-ML-Automation Oct 24, 2025
e4292e6
[Pallas/TPU] Add supports_sparse_core property to TpuInfo.
Google-ML-Automation Oct 24, 2025
2240fb6
[test] handle accelerated dep of lax.dot positional args
jakevdp Oct 24, 2025
cb69cab
[mutable-arrays] in vjp3, don't stuff GradValue sentinels into user p…
mattjj Oct 24, 2025
eeaa805
[mutable-arrays] enable vjp3 by default
mattjj Oct 24, 2025
adbe62b
Reverts eeaa8050f7488b580e5667d43360c98f71b22a05
Google-ML-Automation Oct 25, 2025
11befd4
Update XLA dependency to use revision http://github.com/openxla/xla/c…
Google-ML-Automation Oct 25, 2025
21027c6
Merge pull request #29630 from carlosgmartin:scan_unroll_zero
Google-ML-Automation Oct 26, 2025
29ac6cf
style: type annotate flatten_util
nstarman Oct 23, 2025
4ef87ac
Add :CAPITarget dependency missing from MLIR LLVM dialect.
hawkinsp Oct 26, 2025
376bc38
[jax] Migrate jaxlib to xla::Future
ezhulenev Oct 27, 2025
5f8d5f5
[state] Added some docs to `DischargeRule` and `PartialDischargeRule`
superbobry Oct 27, 2025
cc1a62b
[Mosaic] Explain TPUv4 hotfix comment
Google-ML-Automation Oct 27, 2025
a38e469
Tag some more TF-using tests as thread unsafe.
hawkinsp Oct 27, 2025
afe84f8
Merge pull request #28201 from nstarman:typing/ravel_pytree
Google-ML-Automation Oct 27, 2025
3d13f3b
[Mosaic GPU][NFC] Define the layout variable on the result, instead o…
dimitar-asenov Oct 27, 2025
489e660
Update XLA dependency to use revision http://github.com/openxla/xla/c…
belitskiy Oct 27, 2025
0c03c5d
[Mosaic] Fold no-op BitcastVregOp.
WindQAQ Oct 27, 2025
2dad03a
Correct the SHA for the XLA archive at the current commit.
belitskiy Oct 27, 2025
811fbbc
Merge pull request #32859 from jakevdp:lax-dot-dep
Google-ML-Automation Oct 27, 2025
3af51a5
Allow `jax.jit` to be used as a decorator factory.
Oct 27, 2025
15eab7f
Add an implementation for SVD using the polar decomposition on CUDA G…
hawkinsp Oct 27, 2025
fa86b06
Use jit decorator factory throughout JAX source
jakevdp Oct 27, 2025
85ec653
[jax] Migrate jaxlib to xla::ffi::Context to get access to C API and …
ezhulenev Oct 27, 2025
ea88ace
Disable incompatible paged attention tests for TPU v7
justinjfu Oct 27, 2025
c64c215
[Pallas][Mosaic GPU] Add docstrings for `plgpu.kernel`.
justinjfu Oct 27, 2025
b6a58ed
[Mosaic] Add `tpu::UnpackSubelementsOp` canonicalization.
WindQAQ Oct 27, 2025
e4d0354
Increase GMM test tolerance
justinjfu Oct 27, 2025
1efc89c
Reverts 79231ecda4c7d0d7e8c42b0efb5ee97553e623f5
pravnar Oct 27, 2025
4274216
Add mlir/extras/types.py to the wheel.
belitskiy Oct 27, 2025
f190a24
[ci] pin explicit commit of upload-artifact
jakevdp Oct 27, 2025
686e502
Merge pull request #32900 from jakevdp:upload-artifact-pin
Google-ML-Automation Oct 27, 2025
543cbac
Fix `jnp.reshape` with out_sharding and explicit mdoe when `order="F"`
yashk2810 Oct 27, 2025
3f6d2e4
Skip python_callback_test_tpu:IOCallbackTest.test_sequence_pjit_io_ca…
belitskiy Oct 27, 2025
466aff5
Set the concrete mesh to empty when executing a default callback
yashk2810 Oct 27, 2025
9bbbed1
**TLDR;** Track unreduced and reduced inside shard_map along with var…
yashk2810 Oct 27, 2025
24686f9
[Take 2][vjp3] Enable vjp3 by default
mattjj Oct 27, 2025
2ac4523
Add array-api copy semantics to dlpack MakePjrtBuffer
pearu Jul 3, 2025
f56e230
[Pallas TPU] Add a primitive stochastic_round
megrez-yliu Oct 27, 2025
be26e45
Temporarily disable XNNPACK for qdwh_test.
alexander-shaposhnikov Oct 27, 2025
7ad6dc8
Fix jax tests with vjp3
yashk2810 Oct 28, 2025
a316384
Update XLA dependency to use revision http://github.com/openxla/xla/c…
Google-ML-Automation Oct 28, 2025
d7890b9
[Mosaic GPU] Replace inline assembly with NVVM dialect intrinsics in …
allanrenucci Oct 28, 2025
f1131ef
[Mosaic GPU][NFC] Sort layout conjuring hints to explore `TiledLayout…
dimitar-asenov Oct 28, 2025
56d21f0
[Mosaic GPU] Infer vector size when creating `WGStridedFragLayout` in…
allanrenucci Oct 28, 2025
e3b6a70
Don't test mixed-mode 4th-order derivatives in nn_test.
hawkinsp Oct 28, 2025
b086ca2
Reverts 1efc89ceef2d7fa66c43ec3f8165f8c13266c9cf
Google-ML-Automation Oct 28, 2025
ec1e65e
Merge pull request #29963 from pearu:pearu/from_dlpack-forced-copy
Google-ML-Automation Oct 28, 2025
bb4eeb5
[pallas:sc] Ignore JAX-level memory spaces in `plsc.kernel`
superbobry Oct 28, 2025
2787193
[Mosaic GPU] Serialize kernels as bytecode.
bchetioui Oct 28, 2025
3d4c323
[Mosaic] Improve `canonicalize_select` performance
Google-ML-Automation Oct 28, 2025
0a7cf2c
Avoid cycle due to holding onto exceptions and tracebacks.
tomhennigan Oct 28, 2025
7d69442
[pallas:triton] Deprecated `pl.max_contiguous` -- it should now be ac…
superbobry Oct 28, 2025
8d13b49
Skip TPU v7 tests expecting the wrong number of devices.
justinjfu Oct 28, 2025
e8e1bad
Consolidate material on debugging NaNs.
emilyfertig Nov 15, 2024
03bb702
Fix a test skip for tpu_paged_attention_kernel_test:PagedAttentionKer…
belitskiy Oct 28, 2025
48d0dec
Disable multiprocess tests on Windows.
emilyfertig Oct 28, 2025
f61f0cd
[pallas:sc] Added SparseCore specific attributes to `TpuInfo`
superbobry Oct 28, 2025
9b0264b
Skip //jax/tests/pallas:tpu_pallas_distributed_test_tpu_7x:Verificati…
belitskiy Oct 28, 2025
6ac65f3
Make all `all_gather` variants and `psum_scatter` variants a no-op if…
yashk2810 Oct 28, 2025
29c2a77
[Mosaic GPU] Partially support slicing a swizzled dimension.
allanrenucci Oct 28, 2025
3fd7e23
Reverts 870aa9b16cc783c7619ed8b940bdfd8a0d29bb2e
yueshengys Oct 28, 2025
812900a
Merge pull request #24989 from emilyfertig:emilyaf-sharp-bits-nans
Google-ML-Automation Oct 28, 2025
34dc703
[Mosaic GPU] Fix `MosaicGpuDialectTCGen05Test.test_tmem_alloc_dealloc`.
allanrenucci Oct 28, 2025
01d4386
[Mosaic GPU] Refactor `Distinct` constraint in equation system.
allanrenucci Oct 28, 2025
5407730
Skip //jax/tests/pallas:tpu_gmm_test_tpu_7x.
belitskiy Oct 28, 2025
be3bd87
Add changelog entry for jit decorator factory change
jakevdp Oct 28, 2025
5267f78
[Pallas] Update test_interpret_remote_dma_asymmetrical_indexer to com…
justinjfu Oct 28, 2025
ef1b712
Merge pull request #32890 from jakevdp:decorator-factory
Google-ML-Automation Oct 28, 2025
a772a42
Set the shard count for the array_extensibility_test to 3.
belitskiy Oct 28, 2025
a80acf9
Merge pull request #32936 from jakevdp:jit-decorator-changelog
Google-ML-Automation Oct 28, 2025
d443fe5
Increase test shard count to prevent gpu_paged_attention test timing …
rdyro Oct 28, 2025
21f0e7f
[Rollback][hijax] add a flag gating whether vmap is a pri...
mattjj Oct 28, 2025
3ebff21
[Mosaic TPU] Enable 8-bit transposes on generation 6+ instead of cano…
yueshengys Oct 28, 2025
768ca8a
[Mosaic] Fix packing factor passed into `getPaddedSources`.
WindQAQ Oct 28, 2025
ec5a608
[pallas] Do not use deprecated pl.{load,store} in tests
superbobry Oct 28, 2025
698f731
Ignore multiprocess tests on Windows (take 2).
emilyfertig Oct 28, 2025
875a413
[pallas:triton] Moved a few Triton-specific tests to the correspondin…
superbobry Oct 28, 2025
3eb2c86
Add `out_sharding` to bernoulli
yashk2810 Oct 28, 2025
d45f648
Allow culprit finder to dynamically call workflows
nitins17 Oct 28, 2025
5efd1c7
dc1
nitins17 Oct 28, 2025
3e3110c
dc2
nitins17 Oct 28, 2025
f5a960e
break build
nitins17 Oct 28, 2025
c37eb48
dc3
nitins17 Oct 28, 2025
0aec6e7
dc4
nitins17 Oct 28, 2025
b86ab24
dc5
nitins17 Oct 28, 2025
3a88132
dc6
nitins17 Oct 28, 2025
bfe7254
Deprecate the undocumented but public jax.cloud_tpu_init module.
hawkinsp Oct 28, 2025
be07a28
Add vary_unreduced_cast and reduced_vary_cast functions. `vary_unredu…
yashk2810 Oct 29, 2025
cf2211c
Add a flag for the coordination service initialization timeout for mu…
emilyfertig Oct 29, 2025
2d545c3
[Pallas] Add output fusion support for reduce_sum/broadcast
sharadmv Oct 29, 2025
f6101c5
Add lax.tile_p
sharadmv Oct 29, 2025
e6a1918
Reverts cf2211c4688a72720d89ce888f2f104c2a626654
Google-ML-Automation Oct 29, 2025
f967e4c
Reverts f6101c5b56427e21bee1ed202b5e99f193f87780
Google-ML-Automation Oct 29, 2025
4cfd54d
Update XLA dependency to use revision http://github.com/openxla/xla/c…
Google-ML-Automation Oct 29, 2025
f1a8961
[pallas] Deprecated `pl.swap`
superbobry Oct 29, 2025
58bbb9a
[Mosaic TPU] Refactor and generalize the load/reshape fusion
apaszke Oct 29, 2025
f3bf855
[Mosaic GPU][NFC] Extract `RegisterLayout` test utility into a separa…
allanrenucci Oct 29, 2025
e20c402
[pallas:sc] Removed an unnecessary check from the tpu.store_vector_id…
superbobry Oct 29, 2025
e3f37d4
[pmap] Mark `PythonPmapTest.testAllToAllSplitAxis` as thread-unsafe.
danielsuo Oct 29, 2025
1133323
Fix infinite loop bug in index_map from get_p pull block spec rule.
jburnim Oct 29, 2025
35f0ab1
Add/correct skips for failing tpu_pallas(_pipeline)?_test_tpu_7x test…
belitskiy Oct 29, 2025
1d07a32
[Mosaic GPU] Use `test_util.RegisterLayout` in layout inference tests.
allanrenucci Oct 29, 2025
da35ee8
Add a flag for the coordination service initialization timeout for mu…
emilyfertig Oct 29, 2025
7cd239b
[pmap] Disable testAllToAllSplitAxis for GPUs.
danielsuo Oct 29, 2025
983c592
[Mosaic GPU] Introduces `WG_SPLAT` as a supported `RegisterLayout` in…
allanrenucci Oct 29, 2025
ac6a41f
[Pallas:SC] Adds support for `VMEM_SHARED` and `HBM` in `plsc.kernel`…
brianwa84 Oct 29, 2025
357e16d
Updated jaxlib stubs by running the stubgen with --exclude-values
superbobry Oct 29, 2025
8889828
[Mosaic GPU] Extend `Divides` constraint to support `RegisterLayout`.
allanrenucci Oct 29, 2025
7461265
[Mosaic GPU][NFC] Add the `Argument` VariableType.
dimitar-asenov Oct 29, 2025
f0bf93b
[Mosaic GPU][NFC] Rename `OperandOrResult` to `ValueSite`.
dimitar-asenov Oct 29, 2025
9f6a1ae
Use more conventional type naming in `_type_promotion_lattice`.
chr1sj0nes Oct 29, 2025
392be5b
[Mosaic] Fix JSON formatting in TPU custom call config.
Google-ML-Automation Oct 29, 2025
19e1d68
[Mosaic GPU] Fix op traversal during layout inference to be truly pre…
dimitar-asenov Oct 29, 2025
aa36471
[Pallas:MGPU] Add Warpgroup lowering for `lax.slice_p`.
allanrenucci Oct 29, 2025
74c6b63
Extend `isGuaranteedDivisible to pass on Add op when both sides are d…
IvyZX Oct 29, 2025
3d9f195
reshape->store into strided_store rewrite
voznesenskym Oct 29, 2025
a4b99ac
[hijax] fix caching logic for newstyle hi primitives
mattjj Oct 29, 2025
84954db
jnp: add mean argument to jnp.std and jnp.var families
arvindsankar Oct 29, 2025
d9ba98c
[JAX] Require IFRT values and executables to have some `xla::ifrt::Us…
hyeontaek Oct 29, 2025
297fc6b
Merge pull request #32972 from mattjj:fix-newstyle-hijax-abstract-eva…
Google-ML-Automation Oct 29, 2025
a3a460f
Use TestCase.enterContext in JaxTestCase.setUp
hawkinsp Oct 29, 2025
533ee6c
Remove dead code
yashk2810 Oct 29, 2025
3ea50d5
[Mosaic GPU][NFC] Always use pre-order traversal.
allanrenucci Oct 29, 2025
55b5aef
Merge pull request #32668 from arvindsankar:jnp-mean
Google-ML-Automation Oct 29, 2025
e7c5fd3
Use copy instead of reuse semantics for CopyToHost, because reuse doe…
Google-ML-Automation Oct 30, 2025
bd0397d
[Mosaic:TPU][Relayouts] Replicated minor large-to-small retiling (one…
tlongeri Oct 30, 2025
415e819
Update XLA dependency to use revision http://github.com/openxla/xla/c…
Google-ML-Automation Oct 30, 2025
0772c77
[pallas:mosaic] Inlined `tpu_info.supports_sparse_core`
superbobry Oct 30, 2025
e62f724
Set __module__ on jax.ShapeDtypeStruct.
hawkinsp Oct 30, 2025
eace208
[Pallas:MGPU] Make `TransposeRef` a frozen dataclass.
allanrenucci Oct 30, 2025
b82ed8f
Remove/fix a couple of old TODOs.
hawkinsp Oct 30, 2025
c0d8bdb
Use significantly fewer processes for Windows nightlies under pytest.
belitskiy Oct 30, 2025
6671ddb
Remove unnecessary static dimension check on the k argument to tri.
hawkinsp Oct 30, 2025
78fce19
Update Bazel version to 7.7.0.
meteorcloudy Oct 30, 2025
619dbe1
Add dynamic grid with smem output test.
WindQAQ Oct 30, 2025
d2ce04b
When we do `traced = jit(f).trace(duck_stuff)`, we only use duck_stuf…
yashk2810 Oct 30, 2025
944567b
remove job dep
nitins17 Oct 30, 2025
a01e725
break build
nitins17 Oct 30, 2025
59dc114
dc1
nitins17 Oct 30, 2025
c7eb7c3
dc2
nitins17 Oct 30, 2025
5f763fe
dc3
nitins17 Oct 30, 2025
267bfa2
dc4
nitins17 Oct 30, 2025
7d256f8
Make MetaTy work with HiJAX. Before this HiTypes (with values) were b…
yashk2810 Oct 30, 2025
473499a
[NFC] Add `name` argument to `pallas_call` primitive.
sharadmv Oct 30, 2025
5b4fb3a
Merge remote-tracking branch 'upstream/main' into srnitin/github-app
nitins17 Oct 30, 2025
1eecbc5
disable non running jobs
nitins17 Oct 30, 2025
1bdacb7
test passing in custom env vars
nitins17 Nov 4, 2025
22f4670
update to the c4a runner
nitins17 Nov 4, 2025
65416cb
update to linux x86 runner
nitins17 Nov 4, 2025
24ce151
update defaults
nitins17 Nov 4, 2025
b800613
break build
nitins17 Nov 5, 2025
dbda4c1
dc1
nitins17 Nov 5, 2025
29a23ce
dc2
nitins17 Nov 5, 2025
17066bb
dc3
nitins17 Nov 5, 2025
fb11c94
fix build and update runner
nitins17 Nov 5, 2025
0eae134
dc1
nitins17 Nov 5, 2025
b97023a
dc2
nitins17 Nov 5, 2025
b411070
break build
nitins17 Nov 5, 2025
711e292
dc3
nitins17 Nov 5, 2025
a1c403d
dc4
nitins17 Nov 5, 2025
ef98d21
dc5
nitins17 Nov 5, 2025
e2c30f5
dc6
nitins17 Nov 5, 2025
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
4 changes: 0 additions & 4 deletions .bazelrc
Original file line number Diff line number Diff line change
Expand Up @@ -390,8 +390,6 @@ common:rbe --spawn_strategy=remote,worker,standalone,local
# Attempt to minimize the amount of data transfer between bazel and the remote
# workers:
common:rbe --remote_download_toplevel
# Speed up repos creation by downloading ".tar" dists from the mirror.
common:rbe --config=use_tar_archive_files
test:rbe --test_env=USER=anon

# RBE configs for Linux x86
Expand All @@ -415,8 +413,6 @@ common:rbe_linux_x86_64 --config=ci_linux_x86_64

common:rbe_linux_x86_64_cuda_common --config=rbe_linux_x86_64_base
common:rbe_linux_x86_64_cuda_common --repo_env=REMOTE_GPU_TESTING=1
# Speed up CUDA repos creation by downloading ".tar" dists from the mirror.
common:rbe_linux_x86_64_cuda_common --config=use_tar_archive_files
# Enable forward compatibility for CUDA builds because RBE docker image doesn't
# have latest CUDA drivers installed.
common:rbe_linux_x86_64_cuda_common --@cuda_driver//:enable_forward_compatibility=true
Expand Down
2 changes: 1 addition & 1 deletion .bazelversion
Original file line number Diff line number Diff line change
@@ -1 +1 @@
7.4.1
7.7.0
5 changes: 5 additions & 0 deletions .github/workflows/asan.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,11 @@ on:
paths:
- '**/workflows/asan.yaml'
permissions: {}

env:
UV_DEFAULT_INDEX: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"
PIP_INDEX_URL: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
asan:
# Don't execute in fork due to runner type
Expand Down
8 changes: 5 additions & 3 deletions .github/workflows/bazel_cuda_h100_b200.yml
Original file line number Diff line number Diff line change
Expand Up @@ -55,8 +55,8 @@ jobs:
done
outputs:
any_changed: ${{ steps.changed-files.outputs.any_changed || 'false' }}
run_tests:
needs: changed_files
run_tests:
needs: changed_files
if: ${{ github.event.repository.fork == false && (github.event_name == 'schedule' || needs.changed_files.outputs.any_changed == 'true' || github.event_name == 'workflow_dispatch' || contains(github.event.pull_request.labels.*.name, 'CI Optional GPU Presubmit')) }}
runs-on: linux-x86-a4-224-b200-1gpu
container: 'us-docker.pkg.dev/ml-oss-artifacts-published/ml-public-container/ml-build:latest'
Expand Down Expand Up @@ -142,4 +142,6 @@ jobs:
//tests:fused_attention_stablehlo_test_gpu \
//tests:gpu_tests \
//tests:python_callback_test_gpu \
//tests:ragged_collective_test_gpu
//tests:ragged_collective_test_gpu \
//tests/multiprocess:gpu_tests \
//jax/experimental/jax2tf/tests/multiprocess:gpu_tests
6 changes: 5 additions & 1 deletion .github/workflows/bazel_test_tpu.yml
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@
# run-tests:
# - Downloads the jaxlib wheel from a GCS bucket if build_jaxlib is false.
# - Downloads the libtpu wheels.
# - Executes the `run_bazel_test_cpu.sh` script, which performs the following actions:
# - Executes the `run_bazel_test_tpu.sh` script, which performs the following actions:
# - Runs the TPU tests with Bazel.
name: CI - Bazel test TPU
on:
Expand Down Expand Up @@ -79,6 +79,10 @@ on:
default: 'false'
type: string
permissions: {}

env:
PIP_INDEX_URL: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
run-bazel-tests:
defaults:
Expand Down
4 changes: 4 additions & 0 deletions .github/workflows/ci-build.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,10 @@ concurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.ref }}
cancel-in-progress: true

env:
UV_DEFAULT_INDEX: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"
PIP_INDEX_URL: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
lint_and_typecheck:
runs-on: ubuntu-latest
Expand Down
4 changes: 4 additions & 0 deletions .github/workflows/cloud-tpu-ci-nightly.yml
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,10 @@ on:
# document and enforce the permissions here.
permissions:
contents: read

env:
UV_DEFAULT_INDEX: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
cloud-tpu-test:
strategy:
Expand Down
32 changes: 32 additions & 0 deletions .github/workflows/culprit_finder.yml
Original file line number Diff line number Diff line change
@@ -0,0 +1,32 @@
name: Culprit Finder Workflow

on:
pull_request:
branches:
- main
workflow_dispatch:
inputs:
workflow-to-debug:
description: "Which workflow should be run?"
type: string
default: ""
required: true
json_vars:
type: string
description: 'A JSON string of key-value pairs that will be passed as environment variables to the workflow. (e.g., {"NAME": "test", "VERSION": "1.2.3"})'
required: false
default: '{"TEST": "my-cli-test"}'

concurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.ref }}

permissions: {}

jobs:
bisect-hello-world:
if: ${{ inputs.workflow-to-debug == 'hello_world.yml' }}
uses: ./.github/workflows/hello_world.yml

name: Bisect Hello World
with:
json_vars: ${{ inputs.json_vars }}
53 changes: 53 additions & 0 deletions .github/workflows/hello_world.yml
Original file line number Diff line number Diff line change
@@ -0,0 +1,53 @@
name: Hello World
on:
pull_request:
branches:
- main
workflow_dispatch: # allows triggering the workflow run manually
workflow_call:
inputs:
json_vars:
type: string
description: 'A JSON string of key-value pairs that will be passed as environment variables to the workflow. (e.g., {"NAME": "test", "VERSION": "1.2.3"})'
required: false
default: ''

permissions: {}

jobs:
build:

runs-on: linux-x86-n2-16
container: 'us-docker.pkg.dev/ml-oss-artifacts-published/ml-public-container/ml-build:latest'
steps:
- name: Checkout code
uses: actions/checkout@v4
- name: Parse and set dynamic env vars
run: |
echo "Setting environment variables from input..."
echo '${{ inputs.json_vars }}' | jq -r 'to_entries[] | "\(.key)=\(.value)"' >> $GITHUB_ENV
- name: Simulate build
run: |
echo $GITHUB_ENV
echo $TEST
echo "Building the project..."
sleep 10
echo "Build complete...."
exit 1

test:
runs-on: linux-x86-n2-16
container: 'us-docker.pkg.dev/ml-oss-artifacts-published/ml-public-container/ml-build:latest'
steps:
- name: Checkout code
uses: actions/checkout@v4
- name: Intermediate step
run: |
echo "Running tests."
sleep 15
echo "Tests complete........"
- name: Simulate testing
run: |
echo "Running tests."
sleep 15
echo "Tests complete."
4 changes: 4 additions & 0 deletions .github/workflows/jax-array-api.yml
Original file line number Diff line number Diff line change
Expand Up @@ -12,6 +12,10 @@ concurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.ref }}
cancel-in-progress: true
permissions: {}

env:
UV_DEFAULT_INDEX: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
build:
runs-on: linux-x86-n4-16
Expand Down
3 changes: 3 additions & 0 deletions .github/workflows/numpy_nightly.yml
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,9 @@ concurrency:
# Don't cancel in-progress jobs for main/release branches.
cancel-in-progress: ${{ !contains(github.ref, 'release/') && github.ref != 'main' }}

env:
UV_DEFAULT_INDEX: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
test-nightly-numpy:
defaults:
Expand Down
3 changes: 3 additions & 0 deletions .github/workflows/oldest_supported_numpy.yml
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,9 @@ concurrency:
# Don't cancel in-progress jobs for main/release branches.
cancel-in-progress: ${{ !contains(github.ref, 'release/') && github.ref != 'main' }}

env:
UV_DEFAULT_INDEX: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
test-oldest-supported-numpy:
if: ${{ github.event.repository.fork == false && !startsWith(github.head_ref, 'release/') }}
Expand Down
17 changes: 14 additions & 3 deletions .github/workflows/pytest_cpu.yml
Original file line number Diff line number Diff line change
Expand Up @@ -36,9 +36,19 @@ on:
type: string
halt-for-connection:
description: 'Should this workflow run wait for a remote connection?'
type: boolean
required: false
default: false
type: string
default: 'no'
max-processes:
description: 'The value to set for `--maxprocesses` when running pytest. Leave empty to use max processes.'
type: string
default: ''

permissions: {}

env:
UV_DEFAULT_INDEX: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"
PIP_INDEX_URL: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
run-tests:
defaults:
Expand All @@ -58,6 +68,7 @@ jobs:
JAXCI_HERMETIC_PYTHON_VERSION: "${{ inputs.python }}"
JAXCI_PYTHON: "python${{ inputs.python }}"
JAXCI_ENABLE_X64: "${{ inputs.enable-x64 }}"
MAX_PROCESSES: ${{ inputs.max-processes }}

steps:
- uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
Expand Down
10 changes: 7 additions & 3 deletions .github/workflows/pytest_cuda.yml
Original file line number Diff line number Diff line change
Expand Up @@ -45,9 +45,13 @@ on:
type: string
halt-for-connection:
description: 'Should this workflow run wait for a remote connection?'
type: boolean
required: false
default: false
type: string
default: 'no'
permissions: {}

env:
UV_DEFAULT_INDEX: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
run-tests:
defaults:
Expand Down
4 changes: 4 additions & 0 deletions .github/workflows/pytest_tpu.yml
Original file line number Diff line number Diff line change
Expand Up @@ -60,6 +60,10 @@ on:
type: string
default: 'no'
permissions: {}

env:
UV_DEFAULT_INDEX: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
run-tests:
defaults:
Expand Down
2 changes: 1 addition & 1 deletion .github/workflows/rocm-ci.yml
Original file line number Diff line number Diff line change
Expand Up @@ -46,7 +46,7 @@ jobs:
dist_docker \
--image-tag $TEST_IMAGE
- name: Archive jax wheels
uses: actions/upload-artifact@v4
uses: actions/upload-artifact@330a01c490aca151604b8cf639adc76d48f6c5d4 # v5.0.0
with:
name: rocm_jax_r${{ env.ROCM_VERSION }}_py${{ env.PYTHON_VERSION }}_id${{ github.run_id }}
path: ${{ env.WORKSPACE_DIR }}/dist/*.whl
Expand Down
5 changes: 5 additions & 0 deletions .github/workflows/tsan.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,11 @@ on:
- '**/workflows/tsan.yaml'
- '**/workflows/tsan-suppressions*.txt'
permissions: {}

env:
UV_DEFAULT_INDEX: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"
PIP_INDEX_URL: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
tsan:
runs-on: linux-x86-n4-64
Expand Down
5 changes: 5 additions & 0 deletions .github/workflows/upstream-nightly.yml
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,11 @@ on:
paths:
- '**workflows/upstream-nightly.yml'
permissions: {}

env:
UV_DEFAULT_INDEX: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"
PIP_INDEX_URL: "https://us-python.pkg.dev/ml-oss-artifacts-published/pypi-mirror/simple"

jobs:
upstream-dev:
runs-on: linux-x86-n4-64
Expand Down
Loading
Loading