strisunshinewentingwang/pytorch

Pull Requests Commits

[maskedtensor] move __torch_function/dispatch__ functions to a map (#85529)

george-qi committed 2 years ago

e3766e98

Add instructions on how to merge a PR (#85280)

ZainRizvi committed 2 years ago

78937489

Add nvprims `rand_like` support for Dropout (#85077)

kevinstephano committed 2 years ago

c7b17d7e

Fake tensor refactorings (#85498)

eellison committed 2 years ago

1e4c8851

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

pytorchmergebot committed 2 years ago

d10de31c

Revert "add amp tests (#85434)"

pytorchmergebot committed 2 years ago

eb570ab7

Revert "Turn on aliasing tests for fake backwards, Fix Batch norm running mean/var decomp aliasing (#85471)"

pytorchmergebot committed 2 years ago

3b195fd3

Made max_pool2d_with_indices_backward_cuda contiguify `indices` (#85493)

zou3519 committed 2 years ago

f371b526

Add support to traverse all python collection objects (#84079)

ejguan committed 2 years ago

ea72a099

Turn on aliasing tests for fake backwards, Fix Batch norm running mean/var decomp aliasing (#85471)

eellison committed 2 years ago

1e92eb80

add amp tests (#85434)

eellison committed 2 years ago

c2f4bbe6

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

eellison committed 2 years ago

78afa0cf

Revert "[mkldnn_matmul] enable mkldnn matmul for aarch64 bf16 devices (#83671)"

pytorchmergebot committed 2 years ago

2e883d46

[Quant][fx] Enable FX static quantization for LSTM (#85068)

andrewor14 committed 2 years ago

034f2b4d

Cast grad_input to half when input_dtype is half in _softmax_backward_data aten decomposition (#85497)

rdspring1 committed 2 years ago

71dddec6

should_check_strides (#85416)

Krovatkin committed 2 years ago

b4f9b682

Make functorch compilable with Py-3.11 (#85054)

malfet committed 2 years ago

d5cabf79

[ShardedTensor] Add `is_floating_point` (#85483)

awgu committed 2 years ago

56c0c0af

[ShardedTensor] Add `is_meta` (#85482)

awgu committed 2 years ago

c8f78d41

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

awgu committed 2 years ago

05d0eb2a

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

pytorchmergebot committed 2 years ago

8602873a

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

awgu committed 2 years ago

cf0de77c

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

weiwangmeta committed 2 years ago

8bd4724f

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

rohan-varma committed 2 years ago

5f6735ea

Remove deprecated torch.lstsq (#70980)

IvanYashchuk committed 2 years ago

539076e2

Disable decomposition registration on Python-3.11 (#85509)

malfet committed 2 years ago

6380016b

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

r-barnes committed 2 years ago

f0869cc8

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

pytorchmergebot committed 2 years ago

f0a084f3

Skip MPS test from generic M1 testsuite (#85500)

malfet committed 2 years ago

46a6a50f

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

clee2000 committed 2 years ago

92a94210

Newer Older