strisunshinewentingwang/pytorch

Pull Requests Commits

Fix tensorpipe test (#55979)

rohan-varma committed 4 years ago

3646fa36

[c10d] Log API usage of monitored barrier (#55265)

rohan-varma committed 4 years ago

09231b5d

[c10d] monitored_barrier: ensure all ranks pass or none do (#55197)

rohan-varma committed 4 years ago

a5290ade

[PyTorch] Change MaybeOwned tests to use intrusive_ptr and Tensor (#55684)

swolchok committed 4 years ago

86368700

[PyTorch] Avoid double indirection in MaybeOwned's borrowed state (#55685)

swolchok committed 4 years ago

cf7c5dca

Clang-format batchnorm.py and distributed.py (#55971)

Yi Wang committed 4 years ago

d398a705

Clang-format ProcessGroupMPI.cpp (#55969)

Yi Wang committed 4 years ago

132f5c1f

[FX] Add memory_format to shape_prop (#55815)

James Reed committed 4 years ago

8bdea14c

Revert D27710107: [pytorch][PR] Update a `batch_first` arg for transformers like GRU and LSTM.

VitalyFedyunin committed 4 years ago

2bf26965

Port reflection_pad1d to structured kernel (#55531)

Zheng Yan committed 4 years ago

a61d91e8

Fix OSS flaky test_destroy_full_group on MPI backend in pytorch_linux_xenial_cuda10_2_cudnn7_py3_multigpu_test environment by adding a barrier and retrying MPI_Comm_create 3 times (#55921)

Yi Wang committed 4 years ago

de5e3b5e

[NCCL] Join work clean up thread before aborting communicators (#55444)

rohan-varma committed 4 years ago

c218ac3b

deterministic code path for gather_backward for dim = 1 (#55573)

yuguo68 committed 4 years ago

8596ac18

Update a `batch_first` arg for transformers like GRU and LSTM. (#55285)

scaomath committed 4 years ago

2237754b

cmake: Enable (s)ccache for nccl builds (#55814)

seemethere committed 4 years ago

b98f011c

Skip testing torch.float16 in test_isnan (#55906)

malfet committed 4 years ago

c47cc30b

.github: Bump disk size for auto-scaled workers (#55955)

seemethere committed 4 years ago

bf8b790b

Add nondeterministic alert for `index_put_` when `accumulate=False` (#55827)

kurtamohler committed 4 years ago

5a45b1b2

refactor prepare_for_backward (#54977)

zhaojuanmao committed 4 years ago

5ffc4e3b

print average duration for caffe2 benchmark

linbinyu committed 4 years ago

6dd1978d

[Pytorch] Only print gradient of a tensor if it requires_grad (#54446)

JacobSzwejbka committed 4 years ago

d1fac54f

Reland #50999 (Added pow() on CPU for float16 & bfloat16) (#55280)

imaginary-person committed 4 years ago

aceceb3d

[PyTorch] Mark borrowed case as C10_LIKELY in MaybeOwned (#55553)

swolchok committed 4 years ago

de53de39

[PyTorch] Allow copy operations on MaybeOwned (#55419)

swolchok committed 4 years ago

ea446ed6

[JIT] Use type cache in erasing shape information (#55828)

eellison committed 4 years ago

bbdb37b9

Increase token count threshold for calling thrust sort in embedding backward (#49913)

Sukru Eryilmaz committed 4 years ago

8f953ef5

[caffe2] constexpr const

stepancheg committed 4 years ago

72b8864b

[TensorExpr] Rename `Tensor::call` to `Tensor::load` to be consistent with `Buf` and `Placeholder`. (#55826)

Mikhail Zolotukhin committed 4 years ago

7ab654af

[TensorExpr] Remove mask field from Load and Store classes. (#55825)

Mikhail Zolotukhin committed 4 years ago

1263448c

[TensorExpr] Unbreak benchmarks. (#55824)

Mikhail Zolotukhin committed 4 years ago

754b0d07

Newer Older