strisunshinewentingwang/pytorch

Pull Requests Commits

Add option to log recomps (#98564)

mlazos committed 1 year ago

f55e72c0

[experiment] More procs in CI (#98098)

clee2000 committed 1 year ago

9fd3eba6

Flip Switch Redux (#98341)

eellison committed 1 year ago

e302f083

Don't use f-strings in logging calls (1/X) (#98591)

ezyang committed 1 year ago

16ec7efa

[better_engineering][multiplatform] Repalce host_info() check with select for default_compiler_flags (#98306)

kkolur76 committed 1 year ago

79e14f8f

Skip nnmodule hook guards by default (#98371)

wconstab committed 1 year ago

390c51bf

[devX] make labels only count their own occurences (#98551)

PaliC committed 1 year ago

46d765c1

Add ephemeral merging label (#98543)

PaliC committed 1 year ago

d06662fb

inductor(CPU): support dynamic shape for onednn fusion path (#97230)

XiaobingSuper committed 1 year ago

d643a00e

[Inductor] Fix bug in lowering.slice_ when negative start out of range (#98517)

yanboliang committed 1 year ago

77d9742c

Revert "Reduce includes of CUDACachingAllocator.h (#97072)"

pytorchmergebot committed 1 year ago

45a2f6b7

Reduce overhead in CUDAGraph Trees (#98529)

eellison committed 1 year ago

5c8fea56

[quant][pt2e] QNNPackQuantizer support annotation for resnet18 (#98507)

jerryzh168 committed 1 year ago

616f50da

refactor(add privateuseone floder in aten/src/ATen): add a PrivateUse… (#98127)

ykddd committed 1 year ago

5a537e29

[pt2][inductor] hardcode autotuning names (#98351)

nmacchioni committed 1 year ago

29608fd2

[vision hash update] update the pinned vision hash (#98367)

pytorchmergebot committed 1 year ago

3d8ead7e

Fix off-by-1 error in dynamo coverage stats (#98558)

jansel committed 1 year ago

1fb8428d

Disable test_torchinductor_dynamic_shapes on ASAN (#98544)

huydhn committed 1 year ago

2161be08

[reland][inductor] Enable CudaWrapperCodeGen for non-AOT mode (#98534)

desertfire committed 1 year ago

152d65ae

Update _linux-test.yml (#98317)

weiwangmeta committed 1 year ago

d4dbdee5

Dont decompose dropout so it can be pattern matched (#97931)

jansel committed 1 year ago

a0a0b0c7

[quantized] Fix return values of _get_name() in quantized ConvTranspose (#97678)

kiszk committed 1 year ago

482f87a7

[inductor][cpp] fix mul for uint8 (#98473)

Valentine233 committed 1 year ago

88208c6f

[Resubmit] Don't crash on retrieveDesyncReport (#98470)

rohan-varma committed 1 year ago

06eaa097

[inductor][easy] use num_stages=1 for reduction (#98524)

shunting314 committed 1 year ago

4adba70c

Fix the missing PATH in mps workflow after #98522 (#98559)

huydhn committed 1 year ago

86cb7f40

Revert "[dynamo 3.11] enable dynamo unittests in 3.11 (#98104)"

pytorchmergebot committed 1 year ago

22411b6f

Add test c10d ucc tests (#88110)

Fuzzkatt committed 1 year ago

481ecffb

[RFC] Add warning about object-based collectives for GPU tensors to docs. (#97702)

rohan-varma committed 1 year ago

8a29afe9

Speed up LossCTC.cu (#97269)

yuantailing committed 1 year ago

eb5da4df

Newer Older