pytorch/xla

Pull Requests Commits

Update xla_docker_build.tf

ManfeiBai committed 1 year ago

Verified 76cacc76

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

ManfeiBai committed 1 year ago

Verified 479559c4

Remove `TPU_C_API` device type (#6435)

will-cromar committed 1 year ago

Verified d686d65b

Properly lower aten::lerp (#6339)

wonjoo-wj committed 1 year ago

Verified d73bd833

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

changm committed 1 year ago

Verified 17161eb6

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

cota committed 1 year ago

Verified 904676be

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

ysiraichi committed 1 year ago

Verified 16b6bd1b

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

ysiraichi committed 1 year ago

Verified d5e1b49e

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

ysiraichi committed 1 year ago

Verified 30591ad8

Fix model skip yaml file not found issue (#6428)

zpcore committed 1 year ago

Verified e2012881

Revert "Disambiguate `benchmarks` directory search. (#6434)" (#6440)

zpcore committed 1 year ago

Verified bc4fee20

Disambiguate `benchmarks` directory search. (#6434)

ysiraichi committed 1 year ago

Verified 599b9256

update README.md doc to 2.2 (#6437)

zpcore committed 1 year ago

Verified ea9c6481

Avoid fallback for avg_pool - (#6409)

qihqi committed 1 year ago

Verified a0bae821

Fix distributed checkpoint errors when PyTorch CUDA is enabled. (#6421)

ysiraichi committed 1 year ago

Verified 85c66f5f

Revert "aggregate.py: add tab format" (#6424)

cota committed 1 year ago

Verified 010b6f09

aggregate.py: add tab format (#6405)

cota committed 1 year ago

Verified 90d9dd1d

[benchmarks] Fix CUDA device check. (#6419)

ysiraichi committed 1 year ago

Verified be44bc6b

Use PyTorch's dynamo benchmark skip-list. (#6416)

ysiraichi committed 1 year ago

Verified 492fe27e

Fix some more core aten ops (#6414)

wonjoo-wj committed 1 year ago

Verified a83be44c

[torchbench] Re-land: Force `llama_v2_7b_16h` to use SGD. (#6402)

ysiraichi committed 1 year ago

Verified 0baf1dae

Move torch import closer to the place it uses (#6412)

alanwaketan committed 1 year ago

Verified ba6a9f5b

nightly.sh: pass --always to git describe --tags (#6406)

cota committed 1 year ago

Verified 9334ebbe

nightly.sh: add flag to set Hugging Face token (#6410)

cota committed 1 year ago

Verified 4fc697cb

Revert: #6368 to fix nightly CI coverage run. (#6369)

yeounoh committed 1 year ago

Verified 48a1ae3d

Point to v2.2.0 release (#6418)

zpcore committed 1 year ago

Verified f66406bb

Update OpenXLA pin to 20240124 (#6390)

vanbasten23 committed 1 year ago

Verified 0aee211d

[Pallas] PoC Integration (#6340)

alanwaketan committed 1 year ago

Verified 56db8f23

[FSDPv2] Add a user guide (#6386)

alanwaketan committed 1 year ago

Verified de5d7647

aggregate.py: allow picking backends to report on (#6392)

cota committed 1 year ago

Verified 2b7f4354

Older