strisunshinewentingwang/pytorch

Pull Requests Commits

Slight fix in error message for check_for_seq_len_1_nested_tensor (#88690)

mikaylagawarecki committed 2 years ago

a02ea655

[Profiler] Restructure inputs and capture TensorLists. (#87825)

Taylor Robie committed 2 years ago

6e6f929b

[Profiler] Handle ABA for TensorImpl* when assigning IDs (#87133)

Taylor Robie committed 2 years ago

e132c45f

[MPS][BE] Code cleanup (#88529)

malfet committed 2 years ago

078c25df

PatternMatcher supports matching list-typed args (#88656)

SherlockNoMad committed 2 years ago

1d82eba9

[inductor] Fix aten.fmod lowering (#88602)

peterbell10 committed 2 years ago

8e2627d4

[torch] Implement aten::native_batch_norm.out for CPU (#88604)

larryliu0820 committed 2 years ago

f556d735

Fix `CUDA_MAX_THREADS_PER_SM` for `sm_87` (#88644)

eqy committed 2 years ago

3e30a9ea

Minor error message improvements on meta functions (#88677)

ezyang committed 2 years ago

6bb7f4f2

Revert "[cuDNN] (re-open) Enable cuDNN Frontend v8 API by Default (#87669)"

pytorchmergebot committed 2 years ago

d98a884b

Run libtorch trunk build on linux.4xlarge (#88683)

malfet committed 2 years ago

5eecfcf5

Most recently used cache management for TorchDynamo (#88076)

zyq8709 committed 2 years ago

eaf4fe3d

Mark as_strided_ as supporting SymInt in C++ (#88674)

ezyang committed 2 years ago

1b5373fc

Revert "torchdynamo support modules() for nn_module (#88023)"

pytorchmergebot committed 2 years ago

dba88776

Support diag_embed.out decomposition (#88671)

ezyang committed 2 years ago

860e354d

Correctly test that dtype/device match in generated .out kernels for composites (#88672)

ezyang committed 2 years ago

3f6a5601

Propagate layout and pin memory in randint to inner constructor (#88673)

ezyang committed 2 years ago

245144a6

torchdynamo support modules() for nn_module (#88023)

ydwu4 committed 2 years ago

96104c7b

Deprecate TypedStorage, its derived classes, and all of their public methods (#85303)

kurtamohler committed 2 years ago

ee28b865

enable scalar reduction with dim=-1 (#88628)

Natalia Gimelshein committed 2 years ago

53ca5ad3

Dynamo DDP accuracy bench uses find_unused_parameters (#88645)

wconstab committed 2 years ago

89c58196

Clean up SymFloat binding to cover all functions (#88370)

albanD committed 2 years ago

fcc28834

Fix categorization of sym_int method (#88369)

albanD committed 2 years ago

6abaa594

Add torch.distributed.DistBackendError exception type, thrown from C10D_NCCL_CHECK (#88134)

H-Huang committed 2 years ago

bc66ddb5

Create _make_alias to preserve the name of a function when creating an alias (#88114)

lezcano committed 2 years ago

1a7c4b0d

nvprims bookend non compute (#88457)

jjsjann123 committed 2 years ago

af09270e

Revive static_runtime_benchmark build and test (#87660)

huydhn committed 2 years ago

8cb5c554

[ci] increase timeout time of ios test app build (#88611)

suo committed 2 years ago

02c1a304

[Autograd] Use in-place input accumulation fast path for dense Tensors. (#88339)

Taylor Robie committed 2 years ago

8f66ae41

Add missing args to DDP constructor in distributed.pyi (#88209)

yhcharles committed 2 years ago

ffb6e689

Newer Older