pytorch/xla

Pull Requests Commits

Another update to libtpu, jax, and jaxlib pins to see if they are now working.

yaoshiang committed 200 days ago

a4ec2d2c

Update pins again

yaoshiang committed 202 days ago

16b88a31

yaoshiang committed 205 days ago

97114ef9

Add cache support for scan_layers (#9297)

iwknow committed 205 days ago

Verified 52994af5

Wrap def_static to enable warning reporting. (#9380)

zhanyong-wan committed 205 days ago

Verified 39251247

Revert "Avoid unnecessary copy in TensorSource (#8849)" (#9379)

jeffhataws committed 205 days ago

Verified efdf117b

Implement prng_key as a mutable array (#9305)

tengyifei committed 205 days ago

Verified 0ee876f3

Include both pytorch and torch_xla revisions in the compilation cache key. (#9383)

zhanyong-wan committed 205 days ago

Verified aaff9591

Add autocast feature as `torchax.amp.autocast`. (#9364)

qihqi committed 205 days ago

Verified 89cc066a

Update the tuned block size. (#9376)

QiliangCui committed 205 days ago

Verified f20b89f9

Deprecate ShapeOfXlaOp in favor of GetShape (#9381)

zhanyong-wan committed 205 days ago

Verified b737b4a1

Fix race condition in runAtenTest (#9306)

benawilson committed 206 days ago

Verified f8bd6c48

Improve the API for binding python functions in C++ (#9351)

zhanyong-wan committed 206 days ago

Verified 86c59725

update l_ref in kernel matrix calculation (#9372)

Chenyaaang committed 206 days ago

Verified 77c40f3f

Avoid recompilation caused by `scan_layers` (#9367)

tengyifei committed 207 days ago

Verified 41b9e452

fix all_gather_into_tensor test and logic (#9332)

bfolie committed 207 days ago

Verified 9b0b02f0

Improve `assume_pure` SPMD functionality (#9360)

tengyifei committed 207 days ago

Verified 0be0fbdf

Add an option to not use dlpack. (#9304)

qihqi committed 207 days ago

Verified 55a75404

docs: fix contribute guide command (#9355)

giacomoni committed 208 days ago

Verified 20899c72

Update ragged paged attention kernel to prevent vmem oom (#9346)

Chenyaaang committed 210 days ago

Verified bea86eec

Split _ops and _decomps. (#9323)

qihqi committed 210 days ago

Verified 3a1ed628

Ignore the MODULE.bazel* files generated by bazel. (#9352)

zhanyong-wan committed 210 days ago

Verified 15366e94

Switch XLA_CHECKs to ABSL_CHECKs in lowering_context. (#9338)

zhanyong-wan committed 210 days ago

Verified c2aa21ee

cleanup: remove defunct --distinct_host_configuration flag (#9300)

rickeylev committed 211 days ago

Verified 1d409399

Fix usage of dimensions_size() to check for tuple (#9347)

bfolie committed 211 days ago

Verified 49b0b3b3

Surface CUDA deprecation warning. (#9333)

ysiraichi committed 211 days ago

Verified edd38caa

Update pins (#9330)

pgmoka committed 211 days ago

Verified 503f8280

doc: update pytorch-on-xla-devices and troubleshoot doc for tensor synchronization issue (#9258)

aws-yyjau committed 211 days ago

Verified 3bc9ece7

Update submodules whenever we switch branches. (#9320)

zhanyong-wan committed 212 days ago

Verified 1bd3042f

Remove nightly CUDA builds (#9329)

tengyifei committed 212 days ago

Verified ed14590a

Older