strisunshinewentingwang/pytorch

Pull Requests Commits

srossross committed 2 years ago

cbfef960

change name to FuncTorchBatchedDecomposition

srossross committed 2 years ago

6d5136ae

srossross committed 2 years ago

1e5ae21b

srossross committed 2 years ago

56676251

srossross committed 2 years ago

82a5c882

srossross committed 2 years ago

1b08a03e

srossross committed 2 years ago

420c46ed

srossross committed 2 years ago

53f694b1

srossross committed 2 years ago

adcfe134

Working locally

srossross committed 2 years ago

0e997782

srossross committed 2 years ago

a919a448

srossross committed 2 years ago

9e90dbb4

Editorial pass on Dyamo docs (#89921)

Svetlana Karslioglu committed 2 years ago

015b05af

[ONNX] Supports scatter_add with different static shape of src and index (#89787)

titaiwangms committed 2 years ago

b2f34055

[Quant][fx][bc-breaking] Rename fx/*patterns.py (#89872)

andrewor14 committed 2 years ago

d8005631

fix citation file in MANIFEST (#89994)

pmeier committed 2 years ago

314e7c37

Remove DDP import (#89982)

rohan-varma committed 2 years ago

a5532929

Add error repro test for FSDP ignored modules with mixed precision (#89971)

mrshenli committed 2 years ago

5a36d998

add vjp test with non-contig inputs (#89375)

kshitij12345 committed 2 years ago

dfb533ca

Type torch._dynamo.guards (#89919)

ezyang committed 2 years ago

99dac4dd

Guarantee symbol allocation for all sizes/strides/storage offset (#89879)

ezyang committed 2 years ago

e03cde07

Add definitely_not_01 set to ShapeEnv. (#89871)

ezyang committed 2 years ago

74bcf2b6

When dealing with dupe arguments, prefer leafifying if possible (#89896)

ezyang committed 2 years ago

8d333761

[FSDP][Dynamo] Define annotation attributes as globals (#89913)

awgu committed 2 years ago

808cb2e8

remove torch.equal usages (#89527)

pmeier committed 2 years ago

4095ef8b

fix assert_close docstring (#89620)

pmeier committed 2 years ago

0acbcef4

document torch.testing.assert_allclose (#89526)

pmeier committed 2 years ago

d72cd4c4

enable ufmt for torch/testing/*.py (#89525)

pmeier committed 2 years ago

4baa78bb

ad more error info for cublasLtMatmul (#89983)

zhaojuanmao committed 2 years ago

850b53bb

Add manual meta implementations to quantize_per_tensor.tensor and co (#89958)

ezyang committed 2 years ago

a7473264

Older