strisunshinewentingwang/pytorch

Pull Requests Commits

Fix broken FX tests (#83187)

datumbox committed 2 years ago

6a09847c

[PyTorch] Solved two syntax issues when dumping execution graph result to json file. (#81854)

mingyul-fb committed 2 years ago

ccfbbc0f

Preserve node's stack trace during retrace (#83050)

SherlockNoMad committed 2 years ago

69156764

[functorch] remove some testing hacks (#83079)

zou3519 committed 2 years ago

3aeb5e4f

Move vmap's OrderedDict pytree support to torch.utils._pytree (#83073)

zou3519 committed 2 years ago

6700a785

Prefer contiguous output from mkldnn_bf16_gemm (#82968)

peterbell10 committed 2 years ago

7d89c3b0

Handle redispatch correctly with tensor subclasses in ProxyTensor mode (#83122)

Chillee committed 2 years ago

66396777

add 3 operators ops (max_pool2d, linear, conv2d)in xnnPack test (#83131)

Cuiqing Li committed 2 years ago

22830c7c

[ROCm] update nightly builds to rocm5.2 (#82353)

jeffdaily committed 2 years ago

2dae93b2

[caffe2] Make ListIterator compatible with libc++ 15 (#83189)

smeenai committed 2 years ago

9590cf6d

[fx] Implement __deepcopy__ for fx.Tracer (#83130)

davidberard98 committed 2 years ago

45e7d026

[ao] Create framework for ModelReport Qconfig Generation (#83091)

vspenubarthi committed 2 years ago

a2ca8933

Add exception handler for stoull in caffe2 (#77557)

apach301 committed 2 years ago

92137519

dispatch API for checking computed table, use it in prim decomps (#82358)

bdhirsh committed 2 years ago

1a51efd8

lift numpy tensor, add randperm support (#83191)

eellison committed 2 years ago

8a6b0761

Update state_dict docs (#83104)

Atcold committed 2 years ago

76953bee

RowwiseMoments: use float as acc type for bfloat16 inputs (#81850)

mingfeima committed 2 years ago

2fe3ea65

fix RowwiseMoments vectorization issue on CPU (#81849)

mingfeima committed 2 years ago

4e9b969b

[ONNX] Update legacy code, initialize onnx_shape_inference=True by default (#82767)

BowenBao committed 2 years ago

017ecb78

[NNC] enable fusion of conv with elementwise OP (#77157)

chunyuan-w committed 2 years ago

693a8dd0

Build nccl single-threaded (#83173)

peterbell10 committed 2 years ago

1c83ec8f

Fix test_cuda_primary_ctx being skipped on CUDA (non-ROCm) (#83182)

lw committed 2 years ago

2801f5cf

Reland "[functorch] add error inputs check to vmap test" (#83106)

zou3519 committed 2 years ago

fd3fdf6b

[torchgen] Fix selective build error on custom namespace (#83141)

larryliu0820 committed 2 years ago

c322fc03

[NVFuser] Upstream push 0809 (#83067)

jjsjann123 committed 2 years ago

df741c58

[PyTorch] Allow const T& access to ListElementReference when possible (#83177)

swolchok committed 2 years ago

ce8716f5

Revert "disable some of the periodic ios jobs b/c flaky (#83103)"

pytorchmergebot committed 2 years ago

d58ced3d

Update shallow_copy_and_detach for nested tensor impls (#83002)

drisspg committed 2 years ago

c5c0dd9b

Deleted cuda graph files that were moved to torchdynamo (#83128)

msaroufim committed 2 years ago

0e8beb7d

Turn off keychain timeout in CircleCI builds of ios/TestApp (#83162)

jmdetloff committed 2 years ago

8a7b4b85

Newer Older