strisunshinewentingwang/pytorch

Pull Requests Commits

Deprecate decorating classes with torch.no_grad and similar (#89522)

albanD committed 1 year ago

347a7d97

Add `torch._dynamo` to docs (#89510)

malfet committed 1 year ago

2de38a07

[PT-D][3/N] Sync TP API change to Pytorch (#89535)

fduwjj committed 1 year ago

de0dee30

Call `symint::sizes()` instead of `sizes()` on convolution error messages. (#89549)

ysiraichi committed 1 year ago

795473ff

[quant] Add support for quantize_per_channel in the reference flow with decomposed tensor (#89270)

jerryzh168 committed 1 year ago

39772a6a

[test_nn] split hooks test from test_nn (#89201)

kshitij12345 committed 1 year ago

c651944f

[test_nn] move init tests from test_nn (#89202)

kshitij12345 committed 1 year ago

dd140fc3

Fix Use-after-Free in qembeddingbag_byte_prepack_out (#84750)

Flamefire committed 1 year ago

7594e043

Symintify `select` (#89326)

nkaretnikov committed 1 year ago

07dd2fe6

[quant] Add dequantize_per_channel in quantized_decomposed op library (#89269)

jerryzh168 committed 1 year ago

29742786

Add crossref debug mode for functionalization, catches stride errors (#89498)

ezyang committed 1 year ago

52669534

[BE] Add more `ssh` instructions (#89516)

malfet committed 1 year ago

fe990c8d

Update CUDA compiler matrix (#86360)

Flamefire committed 1 year ago

5b51ca68

Delete unused variable assignment in _refs/__init__.py (#89538)

kit1980 committed 1 year ago

504570d5

Don't use explain() for --explain; instead read it off the counters (#89518)

ezyang committed 1 year ago

ed325119

Allow Module forward-pre and forward hooks to take kwargs (#89389)

mrshenli committed 1 year ago

f5d18574

Added implementation and tests for MPS Hardswish (#87952)

thomaslin2020 committed 1 year ago

4935b597

[inductor] Use dense masks for indirect indexing (#89524)

anijain2305 committed 1 year ago

1cfd3858

Add limited FSDP correctness to torchdynamo benchmark (#89469)

wconstab committed 1 year ago

26322544

[Inductor] Limit g++12 installation to Linux (#89472)

malfet committed 1 year ago

7f4b4d28

Fix inductor fallback_random for dropout/rand_like (#89515)

wconstab committed 1 year ago

b50699f2

[dashboard] Add metric graphs back to dashboard (#89531)

williamwen42 committed 1 year ago

8bf8e4d7

[test_nn] fix missing class attributes for NNTestCase (#89200)

kshitij12345 committed 1 year ago

ce856cee

[quant] Add quantize_per_channel in quantized_decomposed op library (#89268)

jerryzh168 committed 1 year ago

391b593c

[dashboard] Remove aot_cudagraphs and nvprims_nvfuser (#89514)

anijain2305 committed 1 year ago

5bba783d

[Vulkan][TCC] Add tests for quantize_per_tensor and dequantize (#89496)

manuelcandales committed 1 year ago

ea920a11

[ROCm] Optimize layer norm backward kernel for ROCm (#87635)

hubertlu-tw committed 1 year ago

74e62a1f

Shard windows periodic job more (#89455)

clee2000 committed 1 year ago

00b7d8ef

[dashboard] Add commit date & fix date related issues (#89517)

williamwen42 committed 1 year ago

77d7f2c6

Fix vectorized trigonometric functions for VSX (#86453)

Flamefire committed 1 year ago

177baf36

Newer Older