strisunshinewentingwang/pytorch

Pull Requests Commits

Merge branch 'cuda_stream_jit' of https://github.com/pytorch/pytorch into cuda_stream_jit

nikithamalgifb committed 3 years ago

616503ac

nikithamalgifb committed 3 years ago

91142f77

nikithamalgifb committed 3 years ago

fa3801a4

Adding JIT CUDA Stream support

nikithamalgifb committed 3 years ago

68eb21b7

Revert D24422354: [pytorch][PR] fix-process-group-counter

lw committed 3 years ago

f230245c

Fix incorrect usage of CUDACachingAllocator (#46605)

xwang233 committed 3 years ago

e0fd590e

Fix grammar and spelling errors (#46713)

Ansley Ussery committed 3 years ago

6c5f6346

Check support_as_strided before using empty_strided. (#46746)

Ailing Zhang committed 3 years ago

4fd2cce9

[FBGEMM][Transposed Conv] add transposed conv support for fbgemm backend for 1d, 2d, 3d (#46607)

YazhiGao committed 3 years ago

129279a3

Eliminate narrowing conversion (#46730)

r-barnes committed 3 years ago

8558c0e6

Add nvtx.range() context manager (#42925)

chrish42 committed 3 years ago

511f89ea

Enable softmax and tiny norm FP16 tests on ROCm (#46363)

ashish committed 3 years ago

88e94da5

Add ReplaceNaN benchmark as baseline (#46685)

ShijunK committed 3 years ago

6ae0a7c9

Make add_relu an internal function (#46676)

albanD committed 3 years ago

27e2ea4c

Enable DataParallel to run zero input Module (#46565)

lixinyu committed 3 years ago

870a5a0d

[quant][fx] EmbeddingBag quantization support (#46678)

supriyar committed 3 years ago

842494af

[quant][fx] Embedding quantization support (#46677)

supriyar committed 3 years ago

e34c825b

Clean up use of Flake8 in GitHub CI (#46740)

samestep committed 3 years ago

fe6fb775

[CI][IOS] Add a arm64 ios job for Metal (#46646)

xta0 committed 3 years ago

bf1ea14f

[CI][IOS] Rename the IOS_VERSION (#46645)

xta0 committed 3 years ago

344abd56

ProcessGroupNCCL::alltoall_base needs to call recordStream (#46603)

jeffdaily committed 3 years ago

ce5bca55

[quant][graphmode][fx] Add support for additional_fuse_method_mapping (#46345)

jerryzh168 committed 3 years ago

bd90379d

[pt][static_runtime] Add option enable_out_variant (#46690)

Hao Lu committed 3 years ago

d6519d4e

Remove dilation restriction on cuDNN ConvTranspose2d (#46290)

xwang233 committed 3 years ago

f326f6a8

[caffe2] Fix inplace ops in onnx::SsaRewrite (#46134)

Hao Lu committed 3 years ago

53dff784

[caffe2] Allow memonger to optimize nets with inplace(enforced) ops (#46560)

Hao Lu committed 3 years ago

51bf7bed

[quant][graphmode][fx] Add additional_qat_module_mapping (#46344)

jerryzh168 committed 3 years ago

23fad911

torch.nn.Unfold accepts 0-dim for batch size (#40689)

v0dro committed 3 years ago

982fa07c

Revert "Push rocm to slow path (#46216)" (#46728)

Iurii Zdebskyi committed 3 years ago

c57c5607

[FX] Make wrapped functions traceable (#46692)

James Reed committed 3 years ago

9ccf85b7

Older