strisunshinewentingwang/pytorch

Pull Requests Commits

Increase timeout and retry count conda upload (#85802)

atalman committed 2 years ago

879ae452

Add python `nested_tensor` and `as_nested_tensor` constructors in `torch.nested` (#85593)

mikaylagawarecki committed 2 years ago

afaee00f

Expose torch._will_engine_execute_node (#84773)

soulitzer committed 2 years ago

a876432a

[primTorch] Add ref for `huber_loss` and error inputs (#85041)

nkaretnikov committed 2 years ago

8dd45424

add amp tests (#85434)

eellison committed 2 years ago

0b93afb1

test_decomp.py: Skip tests for embedding_backward bf16 (#84554)

peterbell10 committed 2 years ago

29c78266

TestModule: Don't assume sample inputs version counter is zero (#85734)

peterbell10 committed 2 years ago

c2e9b9ec

Slightly beefed up dynamic shapes tests for storage_offset (#85806)

ezyang committed 2 years ago

5b476e68

[Profiler] Optimizer param_groups (part 3 of Record Optimizer) (#85784)

slgong-fb committed 2 years ago

d7766937

Fix gelu repr (#85790)

anijain2305 committed 2 years ago

2f8cfb74

[FSDP] Add `FSDPExtensions` for TP support (#85039)

awgu committed 2 years ago

ff71f457

Revert "Revert "Symintified mmm/addmm derivative formulas (#85794)"" (#85820)

Chillee committed 2 years ago

a4bd89b2

Registered _like metas (#85793)

Chillee committed 2 years ago

39130ccf

Revert "Allow only one -1 in nested view/reshape (#85691)"

pytorchmergebot committed 2 years ago

fc8ba3a9

Revert "Registered _like metas (#85793)"

pytorchmergebot committed 2 years ago

b44a4a8b

[BE] Do not use VLA (#85800)

malfet committed 2 years ago

4c6dc6a1

[JIT] support freezing modules that don't have a forward method (#85779)

davidberard98 committed 2 years ago

424aad7f

Revert "Update `amax/amin/norm/count_nonzero` signatures with `int[*]? dim` (#83300)"

pytorchmergebot committed 2 years ago

a0b16939

Handling for getitem with boolean in meta, and other improvements (#85807)

ezyang committed 2 years ago

224b689c

Don't make parameters have symbolic shapes (#85809)

ezyang committed 2 years ago

b6885f7d

Augment errors raised in fx.Interpreter with Node info (#85810)

ezyang committed 2 years ago

e63d3a8a

[CUBLAS][CUDA GRAPHS] (re-re-open of #83461) Explicitly set the workspace for cuBLAS handles (#85447)

eqy committed 2 years ago

b04b2fa9

Revert "Symintified mmm/addmm derivative formulas (#85794)"

pytorchmergebot committed 2 years ago

823dc33b

[ROCm] Retry loop implemented to avoid transient memory leak errors (#82607)

alugorey committed 2 years ago

5709c67f

[NN module] speed up _load_from_state_dict (#85743)

zhengwy888 committed 2 years ago

b2311192

[BE] small typo+lint fixes for einsum/sumproduct_pair (#85709)

janeyx99 committed 2 years ago

cef8dfc8

Symintified mmm/addmm derivative formulas (#85794)

Chillee committed 2 years ago

230edd25

Registered _like metas (#85793)

Chillee committed 2 years ago

a4e75ccf

Added symbolic shape testing for AOTAutograd (#85789)

Chillee committed 2 years ago

35fe4abd

skip test TestCompositeComplianceCUDA::test_forward_ad_nn_functional_max_unpool2d_cuda_float32 (#85767)

jeffdaily committed 2 years ago

0b251d98

Newer Older