pytorch/xla

Pull Requests Commits

Update gcc to 10. (#5445)

qihqi committed 2 years ago

Verified 454e916e

Move the C++ test build to CI build job instead of test job (#5442)

JackCaoG committed 2 years ago

Verified 4ef4dd34

Bugfix + add more test for llama (#5439)

qihqi committed 2 years ago

Verified a2a4ff4c

[PJRT] Support `torchrun` with `pjrt://` `init_method` (#5438)

will-cromar committed 2 years ago

Verified 9db8fac4

Update fix_includes.sh (#5441)

aazzolini committed 2 years ago

Verified 5b88b5f8

Remove TPU loading hack (#5435)

will-cromar committed 2 years ago

Verified 709c75a0

Add repo for libcudnn8=8.7.0.84 and CUDA 11.8 (#5425)

mateuszlewko committed 2 years ago

Verified f06c31f5

[PJRT|TPU] Update `test_xla_devices_single_process_all_chips` for expected device number (#5421)

ManfeiBai committed 2 years ago

Verified 2d97d3a1

[SPMD] named partition spec support (#5415)

yeounoh committed 2 years ago

Verified 85453d27

Fix the incorect clone arg condition in dynamo bridge (#5414)

JackCaoG committed 2 years ago

Verified 1835771c

SPMD cross slice-replication using partial_replication sharding (#5411)

yeounoh committed 2 years ago

Verified 6852df1d

Revert "Partially replicate lower-rank tensors (#5409)" (#5412)

yeounoh committed 2 years ago

Verified 628ba698

Partially replicate lower-rank tensors (#5409)

jonb377 committed 2 years ago

Verified 56a6a02a

Fix TPU collective ops test for multi-host TPUs (#5408)

will-cromar committed 2 years ago

Verified 2658c882

Refactor to share code between export_torch_model and save_as_stablehlo (#5388)

qihqi committed 2 years ago

Verified 2971efca

[SPMD] Multi-host batch sharded data loading (#5331)

khatwanimohit committed 2 years ago

Verified f2f877f8

Update the dynamo backend name to `openxla` (#5402)

JackCaoG committed 2 years ago

Verified b199b4dc

Fix ReplicateShardedData for int type (#5404)

JackCaoG committed 2 years ago

Verified 17570e20

[PJRT] Separate collective ops test from TPU runtime test. (#5396)

will-cromar committed 2 years ago

Verified a1f822e2

Run single host multi GPU tests in the CI. (#5387)

vanbasten23 committed 2 years ago

Verified bdceee54

Update dynamo cpu fallback op to aten::_foobar (#5393)

wonjoo-wj committed 2 years ago

Verified 6bb6ab5a

Fix change_id type in coverage uploading (#5394)

lsy323 committed 2 years ago

Verified f5b6f1cf

Add manual seed in multihost checkpoint (#5392)

jonb377 committed 2 years ago

Verified 55b941d2

Add dynamo test in TPU CI (#5381)

ManfeiBai committed 2 years ago

Verified 08794f99

Add support for in-place ops with self tensors in dynamo bridge (#5309)

wonjoo-wj committed 2 years ago

Verified 67ff7ea6

Port resnet data loading optimizations to SPMD test script (#5386)

jonb377 committed 2 years ago

Verified 424d8c8e

clear pending ir should also clear the cc op tokens (#5385)

JackCaoG committed 2 years ago

Verified 2676a200

Add gpu doc for how to build PyTorch/XLA from source with GPU support. (#5384)

vanbasten23 committed 2 years ago

Verified c0a432b9

Rectify test_zero1.py once optim.load_state_dict doesn't guarantee immutability (#5382)

janeyx99 committed 2 years ago

Verified ca5eab87

[Dynamo|TPU] Tweak `atol` and `rtol` for `test_simple_model_with_different_input_shape` on TPU (#5373)

ManfeiBai committed 2 years ago

Verified fbec7c10

Newer Older