strisunshinewentingwang/pytorch

Pull Requests Commits

.github: Bump disk size for auto-scaled workers (#55955)

seemethere committed 3 years ago

bf8b790b

Add nondeterministic alert for `index_put_` when `accumulate=False` (#55827)

kurtamohler committed 3 years ago

5a45b1b2

refactor prepare_for_backward (#54977)

zhaojuanmao committed 3 years ago

5ffc4e3b

print average duration for caffe2 benchmark

linbinyu committed 3 years ago

6dd1978d

[Pytorch] Only print gradient of a tensor if it requires_grad (#54446)

JacobSzwejbka committed 3 years ago

d1fac54f

Reland #50999 (Added pow() on CPU for float16 & bfloat16) (#55280)

imaginary-person committed 3 years ago

aceceb3d

[PyTorch] Mark borrowed case as C10_LIKELY in MaybeOwned (#55553)

swolchok committed 3 years ago

de53de39

[PyTorch] Allow copy operations on MaybeOwned (#55419)

swolchok committed 3 years ago

ea446ed6

[JIT] Use type cache in erasing shape information (#55828)

eellison committed 3 years ago

bbdb37b9

Increase token count threshold for calling thrust sort in embedding backward (#49913)

Sukru Eryilmaz committed 3 years ago

8f953ef5

[caffe2] constexpr const

stepancheg committed 3 years ago

72b8864b

[TensorExpr] Rename `Tensor::call` to `Tensor::load` to be consistent with `Buf` and `Placeholder`. (#55826)

Mikhail Zolotukhin committed 3 years ago

7ab654af

[TensorExpr] Remove mask field from Load and Store classes. (#55825)

Mikhail Zolotukhin committed 3 years ago

1263448c

[TensorExpr] Unbreak benchmarks. (#55824)

Mikhail Zolotukhin committed 3 years ago

754b0d07

[TensorExpr] Redesign Rfactor loopnest transformation. (#55324)

Mikhail Zolotukhin committed 3 years ago

b01a15d3

[TensorExpr] Remove unused `LoopNest::hasLoopBodyFor` method. (#55323)

Mikhail Zolotukhin committed 3 years ago

57f795c2

Use autosummary on torch.fft, torch.linalg (#55748)

mattip committed 3 years ago

f61556a7

[NCCL] Log when barrier guesses device to use (#54991)

rohan-varma committed 3 years ago

657b66e8

[package] Correct usage of miniz API in PyTorchStreamReader (#55725)

suo committed 3 years ago

0517222d

Better types in fbcode/caffe2/torch/jit/_script.py (#55856)

r-barnes committed 3 years ago

c3a49cb3

[RFC]fix test_ddp_logging_data_cpu with tsan (#54465)

zhaojuanmao committed 3 years ago

85b97e44

fix ddp logging flaky test (#55414)

zhaojuanmao committed 3 years ago

2eebd9fd

[ROCM] Enable more dtypes in common_method_invocations (#55808)

arindamroy-eng committed 3 years ago

800fa5f3

[Static runtime] refactor MemoryPlanner codes to prepare for output tensor memory planning (#55809)

Peng Wu committed 3 years ago

18662d43

Add stricter typing to caffe2/torch/distributed/elastic/multiprocessing/errors/__init__.py (#55848)

r-barnes committed 3 years ago

6269efde

Use nodes instead of node

Avinash Nagaraj Bukkittu committed 3 years ago

70a09d97

move logic to skip a redispatch directly inside of resize_output (#55162)

bdhirsh committed 3 years ago

2bb58a06

Fix overflow issue in quantized instance_norm/layer_norm/group_norm (#54872)

leslie-fang-intel committed 3 years ago

87fcf307

Factor out numerical logic (#54479)

soulitzer committed 3 years ago

8c8f8829

Refactor get numerical jacobian to calculate wrt all outputs at once (#54378)

soulitzer committed 3 years ago

381b3d8f

Newer Older