strisunshinewentingwang/pytorch

Pull Requests Commits

Catherine Lee committed 2 years ago

7c996198

Catherine Lee committed 2 years ago

ec3476b6

Enable ruff in lintrunner (#99785)

justinchuby committed 2 years ago

7d2a18da

[MPS] Add PSO caching for advanced indexing kernels (#99855)

DenisVieriu97 committed 2 years ago

dcd686f4

Improve the precision of abs() and sign() for large values (#99550)

lezcano committed 2 years ago

09b189ed

Update channel shuffle to return alias instead of self as-is (#99745)

soulitzer committed 2 years ago

5ee5afb8

[xla hash update] update the pinned xla hash (#99863)

pytorchmergebot committed 2 years ago

ab0a8215

Nicer logs for dynamic shapes (#99277)

voznesenskym committed 2 years ago

466877b6

Revert "Do not assume static by default when exporting (#99554)"

pytorchmergebot committed 2 years ago

d0886f68

Revert "Delete tracing_mode argument to export (#99555)"

pytorchmergebot committed 2 years ago

c83e1f51

Add documentation for `torch._logging.set_logs` (#99219)

kurtamohler committed 2 years ago

1e8cf6ad

Fix flake8 lint errors - part 2 - manual fixes (#99799)

justinchuby committed 2 years ago

6e3cdcad

Fix fake tracing of cross entropy with label smoothing and weight (#99830)

Natalia Gimelshein committed 2 years ago

48d112c4

[inductor] Lower aten.prod (#99484)

peterbell10 committed 2 years ago

7a6c650b

Fix flake8 lint errors reported by ruff - take 2 (#99798)

justinchuby committed 2 years ago

79c9e82e

Properly parenthesize dynamo_dynamic_indices test (#99823)

ezyang committed 2 years ago

dc1c0924

[BE] Update python versions for black formatter config (#99827)

Skylion007 committed 2 years ago

6d5040a1

[MPS][BE] Introduce `LookUpOrCreateCachedGraph` (#99422)

malfet committed 2 years ago

f8c68611

[CI] Pause perf data collection for max-autotune (#99829)

desertfire committed 2 years ago

d29cf184

[MPS] Add encoder coalescing support for native kernels (#99810)

DenisVieriu97 committed 2 years ago

a89d6b0a

[Typo]Summary:Fix a typo in comments (#99824)

Bin1024 committed 2 years ago

2d345616

Improve code to make it more pythonic. (#99720)

jovany-wang committed 2 years ago

716ef2f5

[dynamo] Explicitly fall back to eager with GraphModule with no output for onnx&tvm backends (#99805)

YJ Shi committed 2 years ago

72daadef

fix conv+bn folding issue for mixed dtype (#99696)

XiaobingSuper committed 2 years ago

9b0b31a5

inductor: fix split+cat issue when cat order is not align the split output's order (#99700)

XiaobingSuper committed 2 years ago

1fc4d58f

Propagate mark_dynamic in Dynamo compiled outputs. (#99634)

ezyang committed 2 years ago

ebd47b0e

Allow data size equal to 0 for SegmentReduce (#99733)

brad-mengchi committed 2 years ago

efed5a49

Correct LBFGS tolerance_grad doc string (#99792)

arthurfeeney committed 2 years ago

7a8d0ccd

Preserve mark_dynamic when cloning inputs (#99617)

ezyang committed 2 years ago

f602b3a6

Add memoized_only mode to tensor conversion (#99741)

voznesenskym committed 2 years ago

5e73569a

Older