strisunshinewentingwang/pytorch

Pull Requests Commits

Very limited pow support (#87042)

albanD committed 2 years ago

c21dcffc

[xla hash update] update the pinned xla hash (#87067)

pytorchmergebot committed 2 years ago

37e9e89a

[primTorch] Add a ref for `narrow_copy` (#86748)

nkaretnikov committed 2 years ago

91b3cd0b

[primTorch] Implement NLL loss reference (#81128)

rdspring1 committed 2 years ago

847ded6d

[TorchInductor] enable inplace buffers by default (#87037)

jgong5 committed 2 years ago

78e22890

Make `AdamW`, `NAdam` & `RAdam` differentiable (#86183)

Emilio Castillo committed 2 years ago

1b43883f

[vision hash update] update the pinned vision hash (#87021)

pytorchmergebot committed 2 years ago

364a9973

Reland 3 of Merge more symbolic meta kernels and symint changes from branch (#86795)

albanD committed 2 years ago

3a4c0900

[inductor] Disable parallel compile (#87048)

jansel committed 2 years ago

0379af68

stft: Require return_complex to be passed explicitly for real input (#86724)

peterbell10 committed 2 years ago

3007efda

[torchdynamo] Use ProcessPoolExecutor for triton compiles (#87032)

zdevito committed 2 years ago

2b7236a0

Migrate dynamo CI test shards to torch._dynamo (#87039)

jansel committed 2 years ago

945d333a

[inductor] Move size asserts to C++, fix bug (#87028)

jansel committed 2 years ago

30f6f690

[dynamo] Put printing graph breaks behind a config option (#87026)

jansel committed 2 years ago

d45e99ac

OpInfo: Sample input cleanup (4/n) (#86324)

peterbell10 committed 2 years ago

2a6d37d2

[inductor] Use decomps for unfold (#87025)

jansel committed 2 years ago

5099883f

Add labeler with dynamo/inductor paths to start (#87024)

ezyang committed 2 years ago

8a8cd092

Add triton to CI (#86988)

jansel committed 2 years ago

a0c2a7f2

[Dynamo] More robust pyop support, module properties as args (#87020)

voznesenskym committed 2 years ago

3c320a56

OpInfo: Sample input cleanup (3/n) (#86380)

peterbell10 committed 2 years ago

5d6e8315

Sync changes from `pytorch/torchdynamo` (#87013)

jansel committed 2 years ago

054a2fd6

Fixed partitioner issue with getitem and made metadata a storage more consistent (#87012)

Chillee committed 2 years ago

2c1bc216

[einsum] Fix opt_einsum defaults to be more reasonable (#86985)

janeyx99 committed 2 years ago

91c70154

Support unpacking python dictionary in torch.jit.trace() (#81623)

tangleintel committed 2 years ago

7980ed95

[RFC] Separate CPU offload activation to its own wrapper (#85459)

rohan-varma committed 2 years ago

bdefa260

[quant][docs] Formatting fixes for fx graph mode quantization README (#86914)

jerryzh168 committed 2 years ago

100113b8

[vision hash update] update the pinned vision hash (#86758)

pytorchmergebot committed 2 years ago

f6f1aefb

torchdynamo: add linear pointwise(binary) fusion kernel (#86583)

XiaobingSuper committed 2 years ago

46aaae98

torchdynamo: add convolution pointwise(binary) fusion kernel (#86582)

XiaobingSuper committed 2 years ago

5210fab6

torchdynamo: add convolution pointwise(unary) fusion kernel (#86581)

XiaobingSuper committed 2 years ago

9a7a49b2

Newer Older