strisunshinewentingwang/pytorch

Pull Requests Commits

guard include of x64 intrinsics headers (#83793)

Pallab Bhattacharya committed 2 years ago

08c03c91

Move odd num_head in TransformerEncoder to slow_path (#83483)

zrphercule2 committed 2 years ago

e0f2eba9

Skip pr-sanity-checks with skip-pr-sanity-checks label (#83751)

davidberard98 committed 2 years ago

5a1f6d50

Update cpuinfo to the latest commit (#83620)

huydhn committed 2 years ago

f0ee21fe

Freeze the rest of python docs requirement (#83785)

huydhn committed 2 years ago

b2ddef28

torch.cartesian_prod: add type hints (#81377)

adamjstewart committed 2 years ago

9732a7d8

Add support for partial decompositions in make_fx (#83770)

Chillee committed 2 years ago

0e0af73b

Don't extract tensor metadata from sparse tensors (#83669)

ezyang committed 2 years ago

d5a74efc

Refactor is_X_like, better invariant checking for SymInt overload (#83668)

ezyang committed 2 years ago

329deb97

fix functionalization <> fake tensor mode (#83701)

bdhirsh committed 2 years ago

7fe19c03

reinplacing pass fixes for torchbench + huggingface (#83626)

bdhirsh committed 2 years ago

e9e73638

functionalization: handle models that resize their program inputs (#83542)

bdhirsh committed 2 years ago

cce32c6f

Always allow tensor metadata changes (#83590)

bdhirsh committed 2 years ago

0c24af49

[vulkan][ez] lock cache mutex when purging for ShaderCache (#83738)

SS-JIA committed 2 years ago

a7d8863c

Pin sphinxcontrib.katex to 0.8.6 (#83774)

huydhn committed 2 years ago

155343ef

Make some optimizations to minifier (#83641)

Chillee committed 2 years ago

2efbdbfc

[quant][fx][refactor] Rename qconfig_utils.py to qconfig_mapping_utils.py in torch/ao/quantization/fx (#83369)

jerryzh168 committed 2 years ago

13f42069

[primTorch] Add ref for `new_empty_strided` (#82466)

Nikita Karetnikov committed 2 years ago

1f38225b

Enable pg_nccl to perform vector AllGather for uneven output splits (#83713)

aashaka committed 2 years ago

30742193

[Profiler] Start moving python bindings out of autograd (#82584)

Taylor Robie committed 2 years ago

1fa9a377

Remove duplicate_dequantize_node and remove_extra_dequantize (#83611)

danrecoskie committed 2 years ago

7453019e

[Vulkan] Fix issues in GRU and LSTM (#83722)

manuelcandales committed 2 years ago

da520a43

Avoid using fx.Interpreter in nvfuser executor function (#83607)

IvanYashchuk committed 2 years ago

108a1fb1

[vulkan] Throw std::runtime_error instead of using TORCH_CHECK when creating Vulkan context/runtime fails (#83627)

SS-JIA committed 2 years ago

e0d26ee0

Nvfuser python api patch take 2 (#83684)

jjsjann123 committed 2 years ago

1407e672

Refactor CppSignatureGroup to collect signatures as list. (#83667)

ezyang committed 2 years ago

0ec7fc13

Switch fx.replace_pattern to use new SubgraphMatcher (#83717)

SherlockNoMad committed 2 years ago

03e322c8

Avoid unnecessary copy of pointeeSet in MemoryDAG::setWildcards (#83681)

ezgicicek committed 2 years ago

73652dd1

[functorch] re-classify linalg.eigh in vmap testing (#83614)

zou3519 committed 2 years ago

93eedc51

Move `torch.linalg` opinfos to opinfo.definitions (2/2) (#83554)

peterbell10 committed 2 years ago

8788e92f

Newer Older