strisunshinewentingwang/pytorch

Pull Requests Commits

[JIT] Fix compilation unit reference link in constant object upon load (#65784)

Elias Ellison committed 2 years ago

928a4bba

[DataPipe] Fixes an issue where TarArchiveReader closes stream when read into a buffer (#65877)

NivekT committed 2 years ago

81301575

[RFC] [Modular] Include less headers in vararg_functions.cpp (#65672)

iseeyuan committed 2 years ago

7f87ff18

Update CODEOWNERS for optim (#65773)

albanD committed 2 years ago

ea776fa0

Convert Sampler back to lazily construction (#63646)

ejguan committed 2 years ago

b777d790

[quant] update fused_obs_fake_quant op to accept output_fake_quant argument (#65621)

supriyar committed 2 years ago

4666e3f1

[quant] adding memoryless observers for embeddingbag QAT work (#65699)

HDCharles committed 2 years ago

6d4b93bd

Revert D31132861: Make JIT Aliasing Test Less Brittle

suo committed 2 years ago

de80aff7

[Static Runtime] Disable SigridTransform + ListUnpack fusion when outputs reachable from graph output (#62697)

d1jang committed 2 years ago

4176afc4

[DatePipe] add deprecation warnings for DataPipes that will solely exist in TorchData (#65827)

NivekT committed 2 years ago

edab202a

[JIT] Make output of prim::TupleConstruct alias only with its inputs (#64879)

d1jang committed 2 years ago

cd458fe0

Skip failing tests when LAPACK and MAGMA are not available (#64930)

IvanYashchuk committed 2 years ago

dd354117

Remove "SciPioneer" from PT Distributed code owners (#65862)

Yi Wang committed 2 years ago

2c29ec2a

Revert D31005792: [NCCL] Init dummy NCCL comms in constructor

Mike Ruberry committed 2 years ago

91f8755b

Migrate THCIntegerDivider.cuh to ATen (#65745)

ngimel committed 2 years ago

5349ea92

(torchelastic) make --max_restarts explicit in the quickstart and runner docs (#65838)

Kiuk Chung committed 2 years ago

3900509b

[quant] Add imports to the torch/ao/quantization/__init__.py (#64911)

Zafar Takhirov committed 2 years ago

c7ef620a

Avoid saving self for`softmax` and `log_softmax` (#65242)

soulitzer committed 2 years ago

fb412bdd

fix typo in _sharded_tensor (#65511)

crcrpar committed 2 years ago

768cfaa8

Make JIT Aliasing Test Less Brittle (#65493)

Jiewen Tan committed 2 years ago

9f97c66a

Decouple forward AD checks from backward AD in OpInfo tests and gradcheck (#65040)

soulitzer committed 2 years ago

91611fe1

Stop Win+CUDA-10.2 builds (#65649)

malfet committed 2 years ago

5950240b

[NCCL] Init dummy NCCL comms in constructor (#65173)

rohan-varma committed 2 years ago

2b22a5dd

Remove THCDeviceTensor (#65744)

ngimel committed 2 years ago

ad85b582

slow_conv2d_forward: avoid calling dispatcher in parallel region (#65724)

peterbell10 committed 2 years ago

20374c99

Update Module docstring for Python 3 (#65748)

JackKelly committed 2 years ago

7191dd26

ns for fx: add basic testing on cuda (#65593)

vkuzo committed 2 years ago

8bf0ba54

Migrate THCScanUtils to ATen (#65743)

ngimel committed 2 years ago

0dd1b74a

[PyTorch Edge] Conditionally trim dispatch key set to save heap memory at runtime (#65732)

dhruvbird committed 2 years ago

a84feeea

.github: Bump linux gpu max limit to 100 (#65831)

seemethere committed 2 years ago

7b5d676f

Newer Older