pytorch/xla

Pull Requests Commits

Re-enable AMP test on GPU CI. (#6790)

vanbasten23 committed 2 years ago

Verified 5f858e7b

[Pallas] Support multiple outputs (#6844)

alanwaketan committed 2 years ago

Verified 8f095fc8

Support array attribute in stablehlo composite (#6840)

chunnienc committed 2 years ago

Verified 7bbe9d7a

[SPMD] use UNKNOWN as default sharded data sharding type only for auto-sharding (#6846)

yeounoh committed 2 years ago

Verified c1db8759

update 2.3 rc-6 in build triggers (#6845)

lsy323 committed 2 years ago

Verified cba3fe17

update nightly version number (#6842)

lsy323 committed 2 years ago

Verified f6d6286d

Add a dev image for cuda12.3 (#6835)

bhavya01 committed 2 years ago

Verified 59db0cd4

[Pallas] Set the major-minor layout for inputs and outputs (#6826)

alanwaketan committed 2 years ago

Verified d99ceb6b

remove using_pjrt in xla_graph_executor (#6768)

JackCaoG committed 2 years ago

Verified d6fb5391

Remove references to XRT in build and CI tooling (#6761)

will-cromar committed 2 years ago

Verified f083e105

Disable failed C++ Test (#6823)

lsy323 committed 2 years ago

Verified 2b66e93f

Add experimental MLIR debuginfo writer API (#6799)

chunnienc committed 2 years ago

Verified 64a146d8

Use fsspec for multihost checkpoint (#6818)

jonb377 committed 2 years ago

Verified 1ad6bb45

[Pallas] Support XLA_USE_BF16 (#6817)

alanwaketan committed 2 years ago

Verified 73c31dbe

Revert "Install torchvision in release docker" (#6821)

lsy323 committed 2 years ago

Verified b8a97dfc

Add missing shape metadata for the last replaced fx node (#6805)

lsy323 committed 2 years ago

Verified 899a0fa4

Revert "[Reland]Change ATEN generator argument type to const std::optional<Generator>&" (#6819)

alanwaketan committed 2 years ago

Verified 134f5b62

[Fori_loop|While_loop] Add nested test (#6807)

ManfeiBai committed 2 years ago

Verified 22fe1dc3

Install torchvision in release docker (#6795)

lsy323 committed 2 years ago

Verified 7b9177de

Remove reshape lowering (#6801)

wonjoo-wj committed 2 years ago

Verified a84509b2

[Reland]Change ATEN generator argument type to const std::optional<Generator>& (#6804)

cyyever committed 2 years ago

Verified 84e7feb3

Delete torch_patches/.torch_pin (#6812)

JackCaoG committed 2 years ago

Verified 42ecc295

Make compile count in Dynamo test more general (#6802)

JackCaoG committed 2 years ago

Verified 2e5c3b64

Inline all scalars by default in export path (#6803)

lsy323 committed 2 years ago

Verified f43ac41e

Add missing import for XLA init_method (#6809)

jonb377 committed 2 years ago

Verified 34d17c2d

Update XLA pin (#6797)

yeounoh committed 2 years ago

Verified b0ceb2ba

Add cuda 12.1 nightly wheels for cp310 and cp311. (#6793)

vanbasten23 committed 2 years ago

Verified 7d8972c8

ZeRO1: Add bucketting logic to control the size of tensors for all-gather/reduce-scatter (#6025)

jeffhataws committed 2 years ago

Verified e75677f1

Update the instruction of installing the NVIDIA container toolkit (#6791)

vanbasten23 committed 2 years ago

Verified 782f05d2

[Pallas] Allow setting FlashAttention's causal mask (#6792)

alanwaketan committed 2 years ago

Verified de42834d

Newer Older