pytorch/xla

Pull Requests Commits

Infer output dtype for rsub using generic template (#6473)

lsy323 committed 2 years ago

Verified 9884e05a

Fix some more core aten ops (#6462)

wonjoo-wj committed 2 years ago

Verified e8dfc865

Add more ops to torch_xla2 (#6474)

qihqi committed 2 years ago

Verified 6a0cb712

Wrap constant arg in XLAExportInterpreter (#6460)

lsy323 committed 2 years ago

Verified 5dbbb284

bring TPU CI bakc to green by reinstall torch/torchvision (#6458)

ManfeiBai committed 2 years ago

Verified e6baedd5

Add process group documentation for SPMD (#6469)

jonb377 committed 2 years ago

Verified 732a1c7f

moving torch_xla2 experimental code in-tree (#6463)

qihqi committed 2 years ago

Verified 4814d64c

Support multiple StableHLO Composite outputs (#6295)

chunnienc committed 2 years ago

Verified 535d398b

Fix global_device_count(), local_device_count() for single process on CUDA (#6022)

vanbasten23 committed 2 years ago

Verified 8fc8d57a

Reenable TestPoolingNNDeviceTypeXLA (#6453)

qihqi committed 2 years ago

Verified 3e684095

Force guards when performing boolean operations on XLASymNode (#6433)

ezyang committed 2 years ago

Verified fe604e97

[benchmarks] Add support for AMP execution. (#6447)

ysiraichi committed 2 years ago

Verified ad822bf0

torchbench_model.py: fix YAML skip file path (#6457)

cota committed 2 years ago

Verified 510bb8ea

Unify TORCH_LIBRARY definitions (#6455)

alanwaketan committed 2 years ago

Verified 19923bd1

Disable test_EmbeddingBag_per_sample_weights_and_new_offsets (#6456)

JackCaoG committed 2 years ago

Verified 21653dc9

Fix benchmark process terminaion issue (#6441)

zpcore committed 2 years ago

Verified d1a454fe

Disable TestPoolingNNDeviceTypeXLA temporarily. (#6452)

yeounoh committed 2 years ago

Verified d0a5c34e

follow-up for #6420 (#6450)

ManfeiBai committed 2 years ago

Verified 89f7f43b

Add `benchmarks/dynamo` search option for finding PyTorch skip file. (#6446)

ysiraichi committed 2 years ago

Verified f4849154

Re-instate the missing CPU CI tests and their coverage reports (#6448)

yeounoh committed 2 years ago

Verified 52ca01e7

nightly.sh: also run with non-dynamo (i.e. lazy tensor) XLA (#6432)

cota committed 2 years ago

Verified c154b991

aggregate.py: fix plotting of more than 3 backends (#6449)

cota committed 2 years ago

Verified 800a3254

Enable install `google-cloud-sdk-gke-gcloud-auth-plugin` to enable TPUCI run_e2e_tests (#6420)

ManfeiBai committed 2 years ago

Verified 479559c4

Remove `TPU_C_API` device type (#6435)

will-cromar committed 2 years ago

Verified d686d65b

Properly lower aten::lerp (#6339)

wonjoo-wj committed 2 years ago

Verified d73bd833

Add some helpful debugging notes to TROUBLESHOOTING docs (#6430)

changm committed 2 years ago

Verified 17161eb6

Re-land "aggregate.py: add tab format (#6405)", reverted in #6424 (#6431)

cota committed 2 years ago

Verified 904676be

Move tests that need CUDA to `test_operations.py`. (#6160)

ysiraichi committed 2 years ago

Verified 16b6bd1b

Enable CUDA support for PyTorch on CI. (#6070)

ysiraichi committed 2 years ago

Verified d5e1b49e

Check precision only if on CUDA and not inductor. (#6407)

ysiraichi committed 2 years ago

Verified 30591ad8

Newer Older