strisunshinewentingwang/pytorch

Pull Requests Commits

NNC Dynamic Channels last fixes (#72032)

Elias Ellison committed 2 years ago

27a4d397

[NNC] Add Tests for Dynamic Shape Fusion Change default fusion strategy (#71651)

Elias Ellison committed 2 years ago

59a63756

Refactor PE so fusion specializations are configurable (#71650)

Elias Ellison committed 2 years ago

f1499d6c

[WIP] add explicit dynamic fusion arg (#71173)

Elias Ellison committed 2 years ago

cf1833df

Add logspace test modules (#72052)

tugsbayasgalan committed 2 years ago

e305248a

Added remove_duplicate parameter to `nn.Module` (#39)

Chillee committed 2 years ago

7e821754

Upgrade oneDNN to v2.5.2 (#71546)

yanbing-j committed 2 years ago

4567d5de

Add split_with_sizes converter (#71953)

Shirong Wu committed 2 years ago

c61be5fb

Update linspace and bump version nuymber to 8 (#71486)

tugsbayasgalan committed 2 years ago

b28e6965

Make `get_file_pathnames_from_root` output order deterministic (#70435)

bryant1410 committed 2 years ago

5024c1bc

Automated submodule update: FBGEMM (#72116)

facebook-github-bot committed 2 years ago

2c3ecb43

[pytorch] use cublas lt interface for bias fusion (#71200)

jspark1105 committed 2 years ago

702d375d

CUDA TopK Optimization: use multiple block per slice (#71081)

yueyericardo committed 2 years ago

1a30954f

[PyTorch] Rework stat collection in CUDACachingAllocator (#71669)

swolchok committed 2 years ago

4aade950

[PyTorch] Remove call_once from CUDACachingAllocator (#71668)

swolchok committed 2 years ago

ca2ff12e

[PyTorch] Use a better hash table in CUDACachingAllocator (#71667)

swolchok committed 2 years ago

da0423aa

[SR] Add BlockRunner and handle sub-blocks (#69834)

Mike Iovine committed 2 years ago

4b789df6

Simplify TensorImpl size check and fix error message (#72070)

albanD committed 2 years ago

7bb614fc

[JIT] FuseLinear pass now handles CallFunction("linear", ...) (#61646)

peterbell10 committed 2 years ago

ba8d5f6f

Remove some unnecessary python functional wrappers (#61608)

peterbell10 committed 2 years ago

e8d226cd

[quant][fx] Don't assume bias is a keyword-argument (#71426)

peterbell10 committed 2 years ago

7ea96a72

Use new_empty in dropout (#72078)

Samantha Andow committed 2 years ago

a5e27c45

Automated submodule update: FBGEMM (#72068)

facebook-github-bot committed 2 years ago

44e2b8da

improve quantized error checking for structured kernels (#71928)

bdhirsh committed 2 years ago

58dabebc

Revert "[fix] max_pool1d: composite compliance (#70900)" (#71992)

zou3519 committed 2 years ago

f20fa66f

Fix old GCC ABI check in CMake package config (#72081)

cbalioglu committed 2 years ago

1cc824ef

Revert #62143, the new CUDNN_RNN_ALGO_PERSIST_STATIC_SMALL_H algorithm (#72089)

xwang233 committed 2 years ago

c93d6f90

Split cuda: list cpp files that go in _cu library explicitly (#69082)

peterbell10 committed 2 years ago

bb456d2b

DOC: create 1.12 docs from a tag like v1.12.2rc1 (#71985)

mattip committed 2 years ago

e784808b

Make sure we set GITHUB token in the header for pr-label GHA (#72085)

atalman committed 2 years ago

a319bce5

Newer Older