strisunshinewentingwang/pytorch

Pull Requests Commits

[MPS] Perf fixes. (#81951)

kulinseth committed 2 years ago

355d5da2

Format torch/onnx with ufmt (#82137)

huydhn committed 2 years ago

6ea422dd

Really check GRADLE_OFFLINE (#81954)

kit1980 committed 2 years ago

0c8a1c4d

[ONNX] Export aten::_log_softmax (#81804)

wschin committed 2 years ago

6691ed78

Redirect print messages to `stderr` in `torch.utils.cpp_extension` (#82097)

XuehaiPan committed 2 years ago

e849ed3d

[ONNX] extend add and sub exporter to cover graph non-tensor inputs (#81736)

wschin committed 2 years ago

e32691dc

Skip tests that don't call gradcheck in slow gradcheck CI (#82117)

soulitzer committed 2 years ago

0fcdf936

[resubmit][FX] Fix PyTree unpacking carrying forward type annotations (#81999)

James Reed committed 2 years ago

ce92c1cf

[ao][sparsity] Data Sparsifier Benchmarking: Model quality evaluation of the sparsified DLRM model (#81779)

macandro96 committed 2 years ago

eca21fbd

[ao][sparsity] Data Sparsifier Benchmarking: Evaluating disk savings of DLRM model (#81778)

macandro96 committed 2 years ago

1223e944

Update shallow_copy_and_detach for nested tensor impls to enable nested tensor softmax backward (#81838)

drisspg committed 2 years ago

6697f1e4

[PrimTorch] Add reference for torch.norm (#81765)

lezcano committed 2 years ago

11fe277b

[Vulkan] Implement Stack operator (#81064)

manuelcandales committed 2 years ago

f430e0b4

[nn] add `pop` to sequential and ModuleList (#81601)

khushi-411 committed 2 years ago

050aec18

[reland] make ShardedTensor be a Tensor and nn.Parameter (#82089)

wanchaol committed 2 years ago

7ff121e7

Re-enable iOS simulator tests (#82027)

kit1980 committed 2 years ago

40e99fe6

Fix capitalization (#82112)

ezyang committed 2 years ago

e7920c03

Pass stride overload, not overload packet; add aten.stride.default (#82083)

ezyang committed 2 years ago

563f6c7a

All zero sparse tensors are always coalesced (#82085)

ezyang committed 2 years ago

c88fbae4

[FSDP] Refactor casting of grad to full param dtype (#81574)

awgu committed 2 years ago

57a56623

[ONNX] Export aten::convolution (#81815)

wschin committed 2 years ago

1a9317ca

script to monitor memory + cpu utilization (#82006)

clee2000 committed 2 years ago

6f2a88dd

Add some more SparseMeta support (#82086)

ezyang committed 2 years ago

0e995f88

[nn] Add support for +=, * and *= operations for nn.Sequential objects (#81279)

Ansh Radhakrishnan committed 2 years ago

110cd724

change rocblas.h -> rocblas/rocblas.h (#80849)

acxz committed 2 years ago

28262e1e

[JIT] Add backwards compatibility test for old NonDeterminism ops list in ir.cpp (#82029)

goldenxuett committed 2 years ago

7288ea4e

Change autogradNotImplementedFallback to utilize is_mutable and is_aliasing from FunctionSchema (#81917)

goldenxuett committed 2 years ago

3c20b5be

[JIT] Add is_aliasing method to FunctionSchema (#81916)

goldenxuett committed 2 years ago

eb2ea9a5

[JIT] Fix AliasDB edge case for multi non existing alias sets in output (#81837)

goldenxuett committed 2 years ago

52376779

[JIT] Modify is_nondeterministic to utilize tags in schemaInfo and integrate with ir.cpp (#81836)

goldenxuett committed 2 years ago

fc3555ce

Newer Older