strisunshinewentingwang/pytorch

Pull Requests Commits

Apply clang-format to distributed/c10d folder (#107140)

mrshenli committed 1 year ago

dd631919

fix str splits in single line (#106005)

Borda committed 1 year ago

858b465d

[ci] clean up some multigpu tests, and add funcol test (#107153)

wanchaol committed 1 year ago

759c4995

[thread_pg] fix reduce_scatter to respect different cuda device (#107152)

wanchaol committed 1 year ago

9ae51e3a

[thread_pg] fix all_reduce to respect different cuda device (#107151)

wanchaol committed 1 year ago

4be8fe0f

Correct compile doc string format (#107124)

wangxiyuan committed 1 year ago

50927e25

Typo fix in Nonzero.cu (#107090)

EnaAlogo committed 1 year ago

2f0ca722

[Inductor] Make softshrink composite implicit (#107052)

lezcano committed 1 year ago

2c5f96de

Simplify sign lowering in triton (#107051)

lezcano committed 1 year ago

6d899571

Make hardshrink's decomp composite implicit (#107039)

lezcano committed 1 year ago

3b1254e8

Simplify some decompositions. (#107038)

lezcano committed 1 year ago

45c78804

Introduce memory stacks for free (#106758)

zdevito committed 1 year ago

80988b62

[ROCm] enable test_api (test_libtorch) cpp unit tests (#106712)

hongxiayang committed 1 year ago

df849345

[AOTInductor] Prepare for ProxyExecutor, OSS only change (#107065)

SherlockNoMad committed 1 year ago

1e007d04

[TP][DTensor Perf] Some perf improvement to reduce DTensor CPU overhead (#106524)

fduwjj committed 1 year ago

4a6ca4cc

Upload perf benchmark to Rockset in batch of at most 5000 records (#107095)

huydhn committed 1 year ago

00751772

ExportedProgram.transform now updates graph_signature automatically (#107080)

gmagogsfm committed 1 year ago

8c9b2fe8

improve doc on how to understand dynamo (#106860)

youkaichao committed 1 year ago

05db3d99

[dynamo][easy] Only xfail test_dynamic_shapes_float_guard_dynamic_shapes if z3 is available (#107137)

davidberard98 committed 1 year ago

770a565e

Reland: Remove `set_default_dtype` from nn tests (#107069)

kurtamohler committed 1 year ago

6af6b8f7

[Security] Use github environment for update-commit-hash workflow (#107060)

atalman committed 1 year ago

32f93b1c

[Reland] Upgrade NVTX to NVTX3 (#97582)

cyyever committed 1 year ago

5bbfb962

Add typecasting for gelu backward kernel (#86673) (#106856)

wchen61 committed 1 year ago

461c703e

Extend impl_backward to be usable with torch.library operators (#106817)

zou3519 committed 1 year ago

2932b0bf

Extend impl_backward to handle non-Tensor outputs (#106800)

zou3519 committed 1 year ago

db9a0cf6

[custom_op] Allow constructor to infer more types (#106799)

zou3519 committed 1 year ago

9fcce1ba

Run translation validation on tracing error. (#106645)

ysiraichi committed 1 year ago

d8ad7485

[xla hash update] update the pinned xla hash (#107120)

pytorchupdatebot committed 1 year ago

937cd374

Enable mypy check in torch/_inductor/wrapper_benchmark.py (#106775)

FFFrog committed 1 year ago

2b1058c5

[fbcode] Use FastCat in PT Concat implementation (#106727)

zoranzhao committed 1 year ago

d392963a

Newer Older