strisunshinewentingwang/pytorch

Pull Requests Commits

Turn on memory efficient format for jit pickle files.

qihqi committed 2 years ago

aca55948

Adding output_size to to_padded_tensor (#76640)

mjanderson09 committed 2 years ago

76abbbe3

Nested Tensor Elementwise (#76470)

mjanderson09 committed 2 years ago

4d6b145b

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

Chillee committed 2 years ago

8a3e9255

[EASY] Quick Fix for broken shape function autogen.

Gamrix committed 2 years ago

db21e22b

[iOS][coreml] Add CoreML memory observer Round 2

SS-JIA committed 2 years ago

ccd5fa50

[RF][scuba] add pytorch_operator_stats column for Static Runtime out variant (#76566)

ajyu committed 2 years ago

dcdc7b2f

[PT-D][Sharding] Enable ops needed in the transformer model training (#75374)

fduwjj committed 2 years ago

7c44d560

Update isfinite for complex to avoid overflow

zasdfgbnm committed 2 years ago

47e7b12d

Kill dead code in ScanUtils.cuh

zasdfgbnm committed 2 years ago

2e975fb2

Add `TORCH_CPP_LOG_LEVEL` to the docs

ShawnZhong committed 2 years ago

9c902f47

[PyTorch] Remove dead store in intrusive_ptr dtor

swolchok committed 2 years ago

543eaac4

Upstream `argmax` shape function.

silvasean committed 2 years ago

6b6c63ce

Add atan2 isfinite isinf isnan isneginf isposinf isreal to nvfuser and its frontend

zasdfgbnm committed 2 years ago

92d10dec

Improved matmul tests

lezcano committed 2 years ago

9e34a824

[ROCm] tests set PYTORCH_TESTING_DEVICE_ONLY_FOR="cuda"

jeffdaily committed 2 years ago

f1f29ac8

[LT] Move MakeNode into ir_builder.h

desertfire committed 2 years ago

f8a4780e

[ROCm] default tests use 1 GPU, distributed tests use 2 GPUs

jeffdaily committed 2 years ago

138c944a

[LT] Add a flag to control IR reusing

desertfire committed 2 years ago

65b9778d

[ROCm][GHA] relax docker purge conditions

jeffdaily committed 2 years ago

9bfa7e97

stft: remove non-center overload and python functional wrapper

peterbell10 committed 2 years ago

d23ecbfc

Back out "Back out "[torch deploy] Update deploy.rst with working simple example"" (#76713)

s4ayub committed 2 years ago

3e08b181

[FSDP] Faster dict inversion

awgu committed 2 years ago

a4c14caf

[FSDP] Validate exec order using `compute_device`

awgu committed 2 years ago

7772f702

disable the //:generate-code target in Bazel (#76174)

mikey dagitses committed 2 years ago

401179f2

add support for filtering out Bazel targets from common structure (#76173)

mikey dagitses committed 2 years ago

596c54c6

check parameter k and l

awfeequdng committed 2 years ago

9bcb4de1

move generate-code into shared build structure (#75699)

mikey dagitses committed 2 years ago

eb27c851

Remove deprecated string torch::lazy::BackendDevice constructor (#76506)

wonjoolee95 committed 2 years ago

28dfed96

ns for fx: skip operator.add and operator.mul when shadowing (#76504)

vkuzo committed 2 years ago

e155e258

Newer Older