strisunshinewentingwang/pytorch

Pull Requests Commits

Perform appropriate CUDA stream synchronization in distributed autograd.

pritamdamania committed 3 years ago

45d83621

[PyTorch] Move selected_mobile_ops.h codegen function to tools (#53786)

cccclai committed 3 years ago

1772e26f

[TensorExpr] Reland: "Simplify index expressions constructed in loop flattening. Fixes #51173" (#53861)

huiguoo committed 3 years ago

8737c2a1

Move view handling logic to gen_inplace_or_view_type.py (#53341)

Ailing Zhang committed 3 years ago

aeb3e933

[DDP] add _distributed_rank helper function (#53795)

rohan-varma committed 3 years ago

e09e97eb

[DDP] Fix wrong call to dist.get_rank() (#53793)

rohan-varma committed 3 years ago

0c2fe02e

[NNC] Fixes case where inlining wouldn't work because dim-size was 1. (#53254)

Chillee committed 3 years ago

d4602b7e

Revert D26927500: [libkineto] Log CUPTI errors on libkineto initialization

malfet committed 3 years ago

ce670238

Move common autograd utils functions from gen_variable_type.py to api/autograd.py. (#53340)

Ailing Zhang committed 3 years ago

9f75de27

[libkineto] Log CUPTI errors on libkineto initialization

gdankel committed 3 years ago

cffe9aa6

Support loading a non-DP/DDP model from a DP/DDP state_dict (#53224)

Yi Wang committed 3 years ago

d726ce66

[ROCm] Enable RNN test in test_c10d_spawn.py for ROCm (#52707)

KyleCZH committed 3 years ago

5c2b3d77

Disable TF32 on DDP tests (#52941)

zasdfgbnm committed 3 years ago

dfb5f029

[ROCm] Enable test cases in test_data_parallel.py for ROCm (#52708)

KyleCZH committed 3 years ago

06cf6d37

[PyTorch] Fix broken build caused by keyword missing on Windows (#53562)

larryliu0820 committed 3 years ago

c15d9431

[docs] Add starter content for new TorchScript language reference (#53837)

Meghan Lele committed 3 years ago

b69dd910

Revert D26906509: Adding parallel support for the LLVM backend.

malfet committed 3 years ago

d57ae6c4

Remove the extra ":noindex:" in ddp_comm_hooks.rst (#53855)

Yi Wang committed 3 years ago

8d8a4a06

Remove `join_workers` from Pipeline destructor. (#53433)

pritamdamania committed 3 years ago

5344c3ea

Add note on forwarding arguments in the dispatcher (#53641)

smessmer committed 3 years ago

6da0b94d

Automated submodule update: FBGEMM (#53722)

facebook-github-bot committed 3 years ago

13f63fda

[ROCm] Fix the flaky test test_stream_event_nogil (#53850)

jaglinux committed 3 years ago

ec6a7cac

[ONNX] Update inputs/input_names formatting to avoid ValueError with scriptMethods (#53519) (#53548)

BowenBao committed 3 years ago

b9e900ee

Prevent VS from emitting ambiguous symbol errors (third time) (#53490)

skyline75489 committed 3 years ago

cdac61ec

[Gradient Compression] Update the comment on fp16_compress_hook (#53780)

Yi Wang committed 3 years ago

8016d28c

replace data with data_ptr (#53097)

cyyever committed 3 years ago

14d02517

[wip][Dist Profiling] Enable dist profiling for MPI backend (#52949)

rohan-varma committed 3 years ago

fa980bb2

[caffe2] add a SerializationOptions field for the save operator (#53402)

simpkins committed 3 years ago

7e5ffbfa

Implemented getCodeText(string attr) in llvm/cuda codegen and added python bindings for it - #52974 (#53664)

huiguoo committed 3 years ago

1acced4e

Automated submodule update: tensorpipe (#53810)

facebook-github-bot committed 3 years ago

379f1f1e

Older