strisunshinewentingwang/pytorch

Pull Requests Commits

[ShardedTensor] Add `is_meta` (#85482)

awgu committed 1 year ago

c8f78d41

[FSDP] Make `_ran_pre_backward_hook` check more robust (#85481)

awgu committed 1 year ago

05d0eb2a

[vision hash update] update the pinned vision hash (#85522)

pytorchmergebot committed 1 year ago

8602873a

[Easy][FSDP] Simplify `assert` to `p_assert` (#85479)

awgu committed 1 year ago

cf0de77c

Adding a unit test that would gate PRs and prevent reverts, e.g. #83327 (#85442)

weiwangmeta committed 1 year ago

8bd4724f

[FSDP] Address comments on previous PR (#85490)

rohan-varma committed 1 year ago

5f6735ea

Remove deprecated torch.lstsq (#70980)

IvanYashchuk committed 1 year ago

539076e2

Disable decomposition registration on Python-3.11 (#85509)

malfet committed 1 year ago

6380016b

Make CUDA exceptions unlikely and isolate C10_CUDA_CHECK body (#85256)

r-barnes committed 1 year ago

f0869cc8

Revert "[Profiler] Make `LibKinetoClient::stop()` directly call `ProfilerStateBase::pop` (#83965)"

pytorchmergebot committed 1 year ago

f0a084f3

Skip MPS test from generic M1 testsuite (#85500)

malfet committed 1 year ago

46a6a50f

disable circleci jobs b/c they are flaky (#85508)

clee2000 committed 1 year ago

92a94210

Use fallback approach for nested matmul (#85311)

mikaylagawarecki committed 1 year ago

5e700803

[Static Runtime] Fold linear prepack ops (#85289)

Mike Iovine committed 1 year ago

63c1f2fe

[Static Runtime] Fix aten::index_put list conversions (#85298)

Mike Iovine committed 1 year ago

e4899764

Increase timeout for ProcessGroupGlooTest (#85474)

Flamefire committed 1 year ago

bd854588

Revert "[mta] APEX style Fused Adam (#81705)"

pytorchmergebot committed 1 year ago

e505360e

Delete functorch's monkeypatching (#85430)

zou3519 committed 1 year ago

84843759

Move functorch python bindings to torch/csrc (#85426)

zou3519 committed 1 year ago

5e5c3195

Remove deprecated torch.matrix_rank (#70981)

IvanYashchuk committed 1 year ago

bcf93181

Adding conda retry upload to mitigate connection reset errors (#85407)

atalman committed 1 year ago

e3429769

Exposing native _scaled_dot_product_attention to torch.nn (#85044)

drisspg committed 1 year ago

253ffbf2

[GHF] Create "Core Reviewers" group (#85477)

malfet committed 1 year ago

0d04e548

Disabling the pypi cudnn wheel from uploading temporarily (#85470)

atalman committed 1 year ago

e227e3ec

Revert "Add FakeCrossRef tests for backwards, Fix Layer Norm Backward Decomp (#85417)"

pytorchmergebot committed 1 year ago

5043457a

Apply new symbolic shape strategy to make_fx symbolic mode (#85260)

ezyang committed 1 year ago

9baf6770

[ONNX] Update docs on symbolic registration (#85290)

justinchuby committed 1 year ago

2f50d2f6

Add FakeCrossRef tests for backwards, Fix Layer Norm Backward Decomp (#85417)

eellison committed 1 year ago

9c770839

More SymFloat support (#85411)

ezyang committed 1 year ago

61b4e8a7

[maskedtensor] add basic tests and unary/binary/reduction tests from common_method_invocations (#82841)

george-qi committed 1 year ago

0c46e3ec

Newer Older