strisunshinewentingwang/pytorch

Pull Requests Commits

[Composable API] Match `fully_shard()` comm. schedule with wrapper FSDP (#90387)

awgu committed 1 year ago

21a0e809

[Composable API] Refactor `test_fully_shard.py` to use common models (#90386)

awgu committed 1 year ago

4011597d

[Composable API] Move test models to common file (#90385)

awgu committed 1 year ago

5ca4e95f

[inductor] Use decomposition for _to_copy (#90314)

desertfire committed 1 year ago

3fdb5f2d

Upgrade oneDNN to v2.7.2 (#90051)

yanbing-j committed 1 year ago

dc40b6d0

Add a transform for positive-definite matrices. (#76777)

tillahoffmann committed 1 year ago

b4857814

Remove deprecated call to tf.io.gfile.get_filesystem (#89832)

ppwwyyxx committed 1 year ago

c00b135a

Avoid overflow in tensorboard image summary (#90423)

ppwwyyxx committed 1 year ago

ecd78466

[LTC] Overlap data creation and ir_value setting (#90438)

alanwaketan committed 1 year ago

19787733

[Resubmit] state_dict_pre_hook (#90435)

rohan-varma committed 1 year ago

9c80f136

[pruning][core][feature] Implement prune for structured pruning (#89777)

jcaip committed 1 year ago

de016b37

[LTC] Tweak LazyGraphExecutor for XLA (#90420)

alanwaketan committed 1 year ago

c20d4125

[PT-D][Easy] Reformat the optim code within PTD code base (#90399)

fduwjj committed 1 year ago

1a48ae96

Fix issue 38095 TODO in test_multiprocessing.py (#90335)

kit1980 committed 1 year ago

cbb2d5af

[ONNX] Fix ignored small eps in layer normalization in fp16 (#89869)

titaiwangms committed 1 year ago

06c98e67

Revert "add save and load stats in memory_tracker (#90144)"

pytorchmergebot committed 1 year ago

5f3ca208

Revert "[Inductor] More robust stride and offset extraction from index expressions (#90184)"

pytorchmergebot committed 1 year ago

22a249e4

Clean up dependancy for flatbuffer_loader (#86041)

qihqi committed 1 year ago

25eb7c3a

Always compile tiny graphs with AOTAutograd (#89775)

ezyang committed 1 year ago

37892041

[Checkpoint][2D][6/N] Add optimizer and update default_planner to core distributed (#90212)

wz337 committed 1 year ago

b8b74800

Migrate PyTorch to C++17 (#85969)

malfet committed 1 year ago

36ac095f

[pthreadpool] Set max threadlimit to tsan limit (#89453)

digantdesai committed 1 year ago

f2d95765

Revert "Disable dynamo tracing torchrec.distributed (#90087)" (#90416)

wconstab committed 1 year ago

772b7260

Fix issue 38095 TODO in test_jit_fuser_te.py (#90246)

kit1980 committed 1 year ago

00118f5c

Introduce CUDA Device Assertions Infrastructure (#84609)

r-barnes committed 1 year ago

ad188a22

Fix issue 38095 TODOs in gloo tests (#89985)

kit1980 committed 1 year ago

f99f2395

Support pickle version 4 by adding missing ops (#90223)

atuljangra committed 1 year ago

1ba94b38

Rewrite dynamo cond() handling to not recursively call export (#90286)

ezyang committed 1 year ago

d5c6a746

Type torch._dynamo.side_effects (#90202)

ezyang committed 1 year ago

54d344b0

Convert InstructionTranslatorGraphState and OutputGraphState to NamedTuple (#90186)

ezyang committed 1 year ago

ca5f69ef

Newer Older