strisunshinewentingwang/pytorch

Pull Requests Commits

[nnc] Added micro-benchmark to show perf improvement with cat subgraph optimization (#59581)

navahgar committed 3 years ago

47bbc01e

[jit] Added a tranformation to move consumers of aten::cat to its inputs, in the fused subgraphs (#59580)

navahgar committed 3 years ago

d0c4ace0

[jit] Exported a method to get the supported list of elementwise ops (#60162)

navahgar committed 3 years ago

d4c626a3

[jit] Made a list for element-wise ops. (#59579)

navahgar committed 3 years ago

55755edc

[quant][graphmode][fx][refactor] Change the env map to add dtype as a key (#60054)

jerryzh168 committed 3 years ago

a029422c

Be fix shard inbalance (#60206)

Rong Rong (AI Infra) committed 3 years ago

c0f8cad0

[TensorExpr] Add NNC lowerings for `aten::mean`, `aten::addmm`, and `aten::adaptive_avg_pool2d`. (#59347)

Mikhail Zolotukhin committed 3 years ago

d9e7df70

[TensorExpr] Handle not-specified dtypes and strides. (#59346)

Mikhail Zolotukhin committed 3 years ago

c6bb9409

[JIT] ShapeProp: add missing ops from mobilenet v3. (#59163)

Mikhail Zolotukhin committed 3 years ago

f042455a

TF32 threshold twiddling for tests (#60209)

eqy committed 3 years ago

3870e686

[package] Bring back save_source_file (#59962)

zhxchen17 committed 3 years ago

5f010c06

ns for fx: add API usage logging (#60103)

vkuzo committed 3 years ago

5a451031

[static runtime][fix] resize to the input tensor size for full_like (#60229)

ajyu committed 3 years ago

0baad214

[DDP] Support for multiple backwards (#59359)

rohan-varma committed 3 years ago

d5df274e

Enable xenial-cuda11.1-cudnn8-py3.6-gcc7 in GHA (#60196)

samestep committed 3 years ago

3815a013

remove unused `type: ignore` directives (#60006)

pmeier committed 3 years ago

d5988c5e

Fix Subset of a Subset not sliceable issue (#59513)

TJ-coding committed 3 years ago

7c29ca7f

[reland] Move RPC agents to libtorch (#60170)

lw committed 3 years ago

08ce5eed

[reland] Add some TORCH_API annotations to RPC (#60169)

lw committed 3 years ago

958b881d

[reland] Pass RequestCallback to FaultyPG RPC agent (#60168)

lw committed 3 years ago

83fde5d9

Fix saved variable unpacking version counter (#60195)

albanD committed 3 years ago

8a839c54

Adds an OpInfo note (#57428)

Mike Ruberry committed 3 years ago

5609c2e5

Fix clang-tidy path filtering (#60225)

driazati committed 3 years ago

ecc37184

[hierarchical sharding 5/n] enable table-wise -> col-wise sharding in embedding table lookup

Ruilin Chen committed 3 years ago

38c31168

removed cat, equal, and stack from autocast promote list (#59497)

Patrick Wang committed 3 years ago

8b55e9fe

[Profiler] Fix memory profiler merge issue (#60037)

gdankel committed 3 years ago

faf459f1

updated launch bounds for trilinear 3d (#59999)

Fuzzkatt committed 3 years ago

bcf8752f

DOC Describes behavior for None in module.register_* (#60125)

thomasjpfan committed 3 years ago

7e032f18

.github: Run Windows CUDA build on pull requests (#60215)

seemethere committed 3 years ago

047925da

[torch][segment_reduce] Add support for multi-dimensional input (cuda) (#60018)

Serhat Yilmaz committed 3 years ago

6af5d00e

Older