pytorch/xla

Pull Requests Commits

[Kernel] Update ragged attention block table

yaochengji committed 112 days ago

e23e3d41

Add dtensor placement test (#9458)

jeffhataws committed 113 days ago

Verified cc151112

Update torch compat version to 2.7.1 (#9455)

qihqi committed 113 days ago

Verified 01e579c0

Calculate vmem limit dynamically in the quantized matmul kernel. (#9470)

vanbasten23 committed 113 days ago

Verified 83dc9da8

Remove duplicate artifact creation for 2.8.0-rc1 (#9471)

pgmoka committed 113 days ago

Verified 13403086

Update cuda version check (#9469)

pgmoka committed 113 days ago

Verified d450ab37

Fix status source code location logic. (#9440)

ysiraichi committed 114 days ago

Verified 6acc8f3f

Error Handling: replace `XLA_CHECK_OK()` with status functions. (#9457)

ysiraichi committed 114 days ago

Verified 12e77e72

Suppress C++ stacktrace on `XLA_CHECK*()` calls. (#9448)

ysiraichi committed 114 days ago

Verified 5496a364

Error Handling: refactor the PjRt registry to use status QOL functions. (#9419)

ysiraichi committed 114 days ago

Verified 1d848e76

By default, to("jax") should go to TPU (#9468)

zzzwen committed 114 days ago

Verified 1c926f72

Remove the clamp op when we do symmetric quantization on a tensor (#9465)

vanbasten23 committed 114 days ago

Verified cf156c68

Unify the return type of w8a8 matmul between fallback and the actual impl. (#9452)

vanbasten23 committed 114 days ago

Verified 52569ecc

Support editable install with setuptools>=80.0.0 (#9428)

tengyifei committed 114 days ago

Verified 156d9134

Make assume_pure able to work with functions that depends on random (#9460)

qihqi committed 114 days ago

Verified 8999ba55

Change nightly_package_version to 2.9 (#9461)

pgmoka committed 114 days ago

Verified 1ccfede9

Partially disable tpu-info CLI tests (#9463)

bhavya01 committed 114 days ago

Verified 33e87915

Fix CPU tests for python 3.12 (#9443)

bhavya01 committed 114 days ago

Verified 5d05f0a3

Change update_deps script so that latest stable version can be pulled instead of latest nightly (#9424)

bfolie committed 116 days ago

Verified 66ad14f9

Update TPU CI container image (#9434)

bhavya01 committed 121 days ago

Verified 1c00dea2

Add JAX dependency for Python 3.12 (#9438)

tengyifei committed 121 days ago

Verified 53ea5862

Error Handling: refactor `XlaCoordinator` to use status types. (#9386)

ysiraichi committed 122 days ago

Verified 5b8ddb12

Add Build Trigger for 2.8-rc1 release (#9425)

pgmoka committed 122 days ago

Verified 752ddba7

Update CI docker images to use 3.12 (#9427)

bhavya01 committed 123 days ago

Verified 773ecdfb

Optimize w8a8 quantized matmul kernel (#9412)

vanbasten23 committed 123 days ago

Verified 4101ea56

Install libtpu directly instead of torch_xla[tpu] (#9423)

bhavya01 committed 123 days ago

Verified 9c8ae9f9

`EmbeddingDenseBackward`: Remove `padding_idx` cast to `double` (#9406)

unterumarmung committed 124 days ago

Verified 3d92aa75

s/torch_xla2/torchax (#9353)

tengyifei committed 125 days ago

Verified 9499e6f4

implement diagonal_copy (#9416)

Chenyaaang committed 126 days ago

Verified ffce7fb8

Pin update to 20250617. After that there is a pallas regression (#9415)

qihqi committed 126 days ago

Verified 1da73090

Older