strisunshinewentingwang/pytorch

Pull Requests Commits

[PyTorch] Make tls_local_dispatch_key_set inlineable (#49264)

swolchok committed 3 years ago

be849ed1

[CUDA graphs] Cuda RNG-safe graph capture and replay bindings (#48875)

mcarilli committed 3 years ago

c068180a

[CrashFix] Make the dst tensor contiguous when copying from metal

xta0 committed 3 years ago

25833e5d

[AARCH64] Fix vst1q_f32_x2 implementation (#49273)

Sebastian Pop committed 3 years ago

a0432a70

CUDA BF16 sparse (#48807)

zasdfgbnm committed 3 years ago

87636c07

add channels last for AdaptiveAvgPool2d (#48916)

mingfeima committed 3 years ago

690eaf9c

Fix cvtfp32_bf16 (#41280)

malfet committed 3 years ago

8397a62a

Update docstrings of torch.nn.modules.activation.MultiheadAttention (#48775)

Kyeongpil Kang committed 3 years ago

bd322c89

[PyTorch] Make TORCH_CHECK less likely to interfere with inlining (#49263)

swolchok committed 3 years ago

7d406b4a

[PyTorch] native_cpp_binding for size() and stride() (#49262)

swolchok committed 3 years ago

eb051afa

Revert "Revert D25003113: make validate debug-only in Device copy ctr" (#49123)

bdhirsh committed 3 years ago

f54ab8fb

Remove unused operator at::_fft_with_size (#48905)

peterbell10 committed 3 years ago

94a3d4b0

Fix formatting error in `set_deterministic` documentation (#49136)

kurtamohler committed 3 years ago

fdadfb6e

[fx] Add constant folding pass (#48443)

jfix71 committed 3 years ago

38ed3985

Use group.WORLD appropriately in process group initialization. (#48767)

pritamdamania committed 3 years ago

f2ba3c16

Update pipeline API to accept arbitrary sequence of Tensors and not just Tuple (#48467)

pritamdamania committed 3 years ago

dc4db955

fix slow windows test (#49258)

agolynski committed 3 years ago

33b7970d

[pt] Replace size(dim) with sizes()[dim] (#49255)

Hao Lu committed 3 years ago

cd927875

Remove unused reconstruct_scopes function (#48822)

cccclai committed 3 years ago

717f31d9

[te] Use c10::ScalarType utility functions in te::Dtype (#49148)

bertmaher committed 3 years ago

dc92f25b

[te] Use Dtype::is_signed instead of an ad hoc local predicate. (#49147)

bertmaher committed 3 years ago

eaac2819

[te] Fix clamp with uint8 args (#49143)

bertmaher committed 3 years ago

ae88d25c

Fix "Missing return statement" mypy error (#49276)

malfet committed 3 years ago

8999915a

Revert D25434956: [JIT] Use `is_buffer` in `BufferPolicy::valid`

xw285cornell committed 3 years ago

b5b8fe98

[shape inference] fix ConstantFill

ChunliF committed 3 years ago

693e9086

[PyTorch] Remove native::zeros reference in TensorIndexing (#49117)

iseeyuan committed 3 years ago

8d58362f

Enable LayerNorm test cases

venkatacrc committed 3 years ago

635f1cd1

[JIT] Fix toIValue handling of AttributeError when casting ClassType (#49188)

James Reed committed 3 years ago

76d41c80

[Gradient Compression] Minor update of the comments on PowerSGD. (#49246)

Yi Wang committed 3 years ago

29f0fa36

Additional validation for DistributedSampler. (#48865)

pritamdamania committed 3 years ago

21c38e17

Newer Older