strisunshinewentingwang/pytorch

Pull Requests Commits

Adding Peter's Swish Op ULP analysis. (#42573)

venkatacrc committed 4 years ago

e95fbaab

[NCCL] DDP communication hook: getFuture() without cudaStreamAddCallback (#42335)

sinannasir committed 4 years ago

0a804be4

[caffe2] Fix the timeout (stuck) issues of dedup SparseAdagrad C2 kernel

jianyuh committed 4 years ago

d4a4c62d

[fbgemm] use new more general depthwise 3d conv interface (#42697)

jspark1105 committed 4 years ago

3fa0581c

Fix lite trainer unit test submodule registration (#42714)

ann-ss committed 4 years ago

13bc5428

Add fake quantize operator that works in backward pass (#40532)

durumu committed 4 years ago

48e978ba

Exposing Percentile Caffe2 Operator in PyTorch

Edson Romero committed 4 years ago

2b047122

Skips some complex tests on ROCm (#42759)

Mike Ruberry committed 4 years ago

55b17067

Restrict conversion to SmallVector (#42694)

smessmer committed 4 years ago

95f4f675

fix celu in quantized benchmark (#42756)

vkuzo committed 4 years ago

faca3c43

Automated submodule update: FBGEMM (#42713)

facebook-github-bot committed 4 years ago

4eb66b81

[caffe2] add type annotations for caffe2.distributed.python

simpkins committed 4 years ago

02f58bdb

BAND, BOR and BXOR for NCCL (all_)reduce should throw runtime errors (#42669)

thinking-tower committed 4 years ago

6ebc0504

Speed up HistogramObserver by vectorizing critical path (#41041)

durumu committed 4 years ago

7332c21f

C++ API TransformerEncoderLayer (#42633)

lixinyu committed 4 years ago

98de1503

[JIT] Exclude staticmethods from TS class compilation (#42611)

Meghan Lele committed 4 years ago

eba35025

Minor typo fix (#42731)

vfdev-5 committed 4 years ago

9f88bcb5

[vulkan] Fix warnings: static_cast, remove unused (#42195)

IvanKobzarev committed 4 years ago

04c62d4a

Remove duplicate definitions of CppTypeToScalarType (#42640)

peterbell10 committed 4 years ago

586399c0

[NNC] Remove VarBinding and go back to Let stmts (#42634)

nickgg committed 4 years ago

944ac133

Handle fused scale and bias in fake fp16 layernorm

tracelogfb committed 4 years ago

2971bc23

Fix some linking rules to allow path with whitespaces (#42718)

malfet committed 4 years ago

dcee8933

Adds torch.linalg namespace (#42664)

Mike Ruberry committed 4 years ago

9c8021c0

[CPU] Added torch.bmm for complex tensors (#42383)

anjali411 committed 4 years ago

c9346ad3

Fix cmake warning (#42707)

malfet committed 4 years ago

31ed4689

[vulkan] Ops registration to TORCH_LIBRARY_IMPL (#42194)

IvanKobzarev committed 4 years ago

3c66a379

Blacklist to Blocklist in onnxifi_transformer (#42590)

DeepakVelmurugan committed 4 years ago

4eb02add

Add use_glow_aot, and include ONNX again as a backend for onnxifiGlow (#4787)

jfix71 committed 4 years ago

fb8aa004

Updates alias pattern (and torch.absolute to use it) (#42586)

Mike Ruberry committed 4 years ago

73642d94

[blob reorder] Seperate user embeddings and ad embeddings in large model loading script

ChunliF committed 4 years ago

cb1ac940

Newer Older