strisunshinewentingwang/pytorch

Pull Requests Commits

Making cuda 11.5 workflows periodic (#69323)

atalman committed 2 years ago

47bed458

Add conda-forge dependency for cuda-11.5 (#69541)

malfet committed 2 years ago

ccfc76ac

Add Union type to TorchScript Language Ref (#69514)

gmagogsfm committed 2 years ago

215becf3

[JIT] optimize_for_inference on methods other than forward (#69367)

davidberard98 committed 2 years ago

f66ef805

[JIT] run frozen optimizations on methods other than forward (#68668)

davidberard98 committed 2 years ago

fe9be303

Sparse CSR: Implement unary ufuncs (with 0->0 correspondence) (#69292)

krshrimali committed 2 years ago

7b9c82ab

Docs: Fix broken code syntax in autograd.rst (#69362)

rodrigobdz committed 2 years ago

807ad67b

trt engine inspector demo (#66683)

Yinghai Lu committed 2 years ago

8ca1674c

[JIT] check stack size after calling operator (#68788)

davidberard98 committed 2 years ago

678ff498

[DataPipe] Small change to generation script and update to DataPipe .pyi file (#69392)

NivekT committed 2 years ago

d0cab573

[DataPipe] Removing usage of unbatch_level from .batch interface and DataFrame (#69393)

NivekT committed 2 years ago

bbb85999

[DataPipe] Unifying API - removing nesting_level argument from FilterIterDataPipe (#69391)

NivekT committed 2 years ago

9e9c4c80

[DataPipe] Unifying API - removing nesting_level argument from MapperIterDataPipe (#69390)

NivekT committed 2 years ago

d36bd595

[quant][refactor] Move pattern type definition to ao/quantization/utils.py (#68769)

jerryzh168 committed 2 years ago

e1c3793e

[iOS] Add LibTorch nightly build (#69341)

Hanton Yang committed 2 years ago

ccc73256

[PyTorch] Avoid no-op shared_ptr dtor when constructing tuple (#69337)

swolchok committed 2 years ago

77cc7fca

GHA: Re enable mac json uploads (#69387)

janeyx99 committed 2 years ago

3afc2ea7

[AutoAccept][Codemod][FBSourceClangFormatLinter] Daily `arc lint --take CLANGFORMAT`

CodemodService FBSourceClangFormatLinterBot committed 2 years ago

6ccbe846

Set non-default backend names to lower case (#69400)

bryanmr committed 2 years ago

333e6890

FX: ensure node stack trace survives copying (#69368)

vkuzo committed 2 years ago

fa95b301

[quant][graphmode][fx] Add default_replay_qconfig for ops like reshape (#69249)

jerryzh168 committed 2 years ago

0fb557b7

[JIT] Separate GPU implementation of frozen_conv_add_relu_fusion.cpp (#68149)

davidberard98 committed 2 years ago

166df6e1

Enable build-split for all CUDA-11.x version (#69494)

malfet committed 2 years ago

47149a4f

[wip][quant][graphmode] produce reference pattern for binary ops and then rewrite to quantized op (#68229)

jerryzh168 committed 2 years ago

3e7b9494

[fx2trt][easy] add sparse flag to TRTInterpreter (#69495)

842974287 committed 2 years ago

fd6ad018

[fx2trt] Separate sign from `trunc_div` and use it for acc_ops.sign (#69486)

842974287 committed 2 years ago

155be3cd

[fx2trt] [Prep for release] Add type hints to converters and separate main files (#69458)

842974287 committed 2 years ago

df20d9e2

[fx2trt] Add a helper function to generate specs for dynamic batch size (#69405)

842974287 committed 2 years ago

d7d6260b

[fx2trt] Use weights shape as normalize shape in layer norm (#69401)

842974287 committed 2 years ago

7c7cb94a

Revert D32704467: [Autograd/Checkpoint] Checkpoint implementation without reentrant autograd

suo committed 2 years ago

fa3bf8b3

Newer Older