strisunshinewentingwang/pytorch

Pull Requests Commits

Fix false DeprecationWarning in `Module.state_dict`

lkct committed 3 years ago

b8776e14

[NNC] Lowering function generates the output buffer with the specified stride (#76529)

EikanWang committed 3 years ago

429a80dd

[ROCM] remove rocfft workaround

micmelesse committed 3 years ago

0878ba46

[Static Runtime] Add auto generated unstructured ops (#76398)

huiguoo committed 3 years ago

bcddd4ab

[Static Runtime] [RFC] Codegen support for ops with unstructured kernels (#76203)

huiguoo committed 3 years ago

ca0f2670

[SR] Fuse quantized linear/relu (#75775)

Mike Iovine committed 3 years ago

fc64dbdc

aten::_validate_sparse_compressed_tensor_args ops added to the ALLOW_LISST

b0noI committed 3 years ago

7dc13831

[FSDP] Optim state: ignore params if not in dict

Andrew Gu committed 3 years ago

94d65b05

Fixes prim type promotion and updates type promotion testing

Mike Ruberry committed 3 years ago

b557e102

Correct _validate_sparse_compressed_tensor_args signature

pearu committed 3 years ago

fc1f290e

switch Bazel to the shared generate-code genrule (#75790)

mikey dagitses committed 3 years ago

ac45fb9b

introduce new --gen-dir flag to generate_code and use it in fbcode (#75800)

mikey dagitses committed 3 years ago

096ff0ec

Add type hints for a few random functions/classes

adamjstewart committed 3 years ago

dfde877c

add nested mode to python mode

samdow committed 3 years ago

6779366f

Revert "Make addcmul and addcdiv support different dtypes"

pytorchmergebot committed 3 years ago

68fa6d8f

Revert "Add binary_cross_entropy and trace decomp - fixed _log_softmax/_softmax dtype promotion semantics"

pytorchmergebot committed 3 years ago

ce63c53c

[AutoAccept][Codemod][FBSourceClangFormatLinter] Daily `arc lint --take CLANGFORMAT` (#76802)

CodemodService FBSourceClangFormatLinterBot committed 3 years ago

fa3e0d5f

[primTorch] slice and transpose & etc.

Mike Ruberry committed 3 years ago

ef9f56eb

Revert "Adjust the stubs for PyCharm autocompletion of the Tensor methods."

pytorchmergebot committed 3 years ago

9b97313c

Add docstrings to sparse compressed tensor factory functions

pearu committed 3 years ago

5adf97d4

Support str for compressed sparse tensors

pearu committed 3 years ago

2834726e

Factory functions for sparse CSC, BSR, and BSC tensors

pearu committed 3 years ago

436a7be0

[WIP] sum reference

Natalia Gimelshein committed 3 years ago

c51b53d4

[quant][fx][improvement] Renamed default_affine_fixed_qparams_observer and default_symmetric_fixed_qparams_observer (#76637)

dzdang committed 3 years ago

e2aa28a2

Make addcmul and addcdiv support different dtypes

Nikita Karetnikov committed 3 years ago

4cdec1a7

Adjust the stubs for PyCharm autocompletion of the Tensor methods.

cosine0 committed 3 years ago

f8b0f698

Optimize half conversion for SYCL kernel

cchheennhhaaoo committed 3 years ago

68e012b0

Added gelu decomposition

Chillee committed 3 years ago

ed18181d

Use scatter_reduce to support masked reductions on sparse COO tensors (sum, prod, amin, amax)

mikaylagawarecki committed 3 years ago

fc2a2e8b

[NVFuser] environment variable to turn nvfuser on or off (#76485)

davidberard98 committed 3 years ago

e33f3229

Newer Older