strisunshinewentingwang/pytorch

Pull Requests Commits

Trigger azure pipeline for multi gpu tests (#52490)

skyline75489 committed 3 years ago

f0dafeb0

SyncBatchNorm.forward() to handle optional weight (#54568)

malfet committed 3 years ago

69c5fd1e

Update release notes scripts following runbook update (#54594)

albanD committed 3 years ago

f83668b4

[tensorexpr] Add sliceHead/sliceTail APIs with short parameter list (#55115)

huiguoo committed 3 years ago

967e59e5

[FX] Adds C-level monkeypatching of `torch.randn` so that we can capture it during tracing. (#54060)

Chillee committed 3 years ago

1324b0dd

[static runtime] fix out variant for 4bit embedding bag (#55096)

ajyu committed 3 years ago

0cfd9e88

port copysign to structured kernel (#55040)

RockingJavaBean committed 3 years ago

b880854f

Fixed OpInfo jit tests failing for TensorList inputs (#54954)

heitorschueroff committed 3 years ago

8b02d120

Avoid aggregate initialization for tensorpipe::{Cpu,Cuda}Buffer and tensorpipe::Message::Tensor. (#55136)

Lucas Hosseini committed 3 years ago

9d6a81d1

Revert D27449031: [pytorch][PR] [ROCm] use hiprtc precompiled header

agolynski committed 3 years ago

204ac21b

Replace torch.chain_matmul calls to torch.linalg.multi_dot (#55064)

heitorschueroff committed 3 years ago

30367773

Deprecate torch.chain_matmul in favor of torch.linalg.multi_dot (#53453)

heitorschueroff committed 3 years ago

d98072b0

[Relanding] Implemented torch.linalg.multi_dot (#52859)

heitorschueroff committed 3 years ago

5d68b369

Check exception messages in embedding_bag_proxy unit test

yuhc committed 3 years ago

5a1191d0

[tensorexpr] Add python bindings for TensorExprKernel (#54450)

huiguoo committed 3 years ago

50cb75ed

[PyTorch] Use DimVector for inputs to as_strided that don't grow dim (#55016)

swolchok committed 3 years ago

ba95e08a

[caffe2] Reintroduce Log1p operator (#55073)

khabinov committed 3 years ago

6145ac07

[caffe2] Fix -Wundef

stepancheg committed 3 years ago

547346d6

[Gradient Compression] Report compression rate for batched PowerSGD hook (#55103)

Yi Wang committed 3 years ago

058357a4

[PyTorch] Check is_same instead of data_ptr in addmm_out_cuda_impl (#55111)

swolchok committed 3 years ago

d2aab53d

[ONNX] Fix export of copy_ operator (#51938) (#54870)

KsenijaS committed 3 years ago

7824d827

Add outer export to onnx (#53603) (#54869)

hwangdeyu committed 3 years ago

40869884

[ONNX] Update scripting docs (#54634) (#54868)

neginraoof committed 3 years ago

c5f3d928

[ONNX] Replace decomposeLinear pre process pass with a symbolic (#53077) (#54866)

KsenijaS committed 3 years ago

cb0cee4a

[ONNX] Fix if output shape mismatch error & Fix graph input directly used as output (#53219) (#54865)

BowenBao committed 3 years ago

849dcb8b

[ONNX] Support primitive type input/outputs and attributes (#53550) (#54864)

neginraoof committed 3 years ago

cd9dd653

[ONNX] Improve index_put symbolic to handle singular Bool updates (#53690) (#54863)

shubhambhokare1 committed 3 years ago

ce48b140

Allow `std=0` in `torch.normal`, and error if `std<0` (#51317)

kurtamohler committed 3 years ago

6c235ef2

Revert D27408378: [quant][graphmode][fx][refactor] Factor out insert_observers_for_model to a separate function

Mike Ruberry committed 3 years ago

15f04e34

Added OpInfo gradcheck_wrapper to replace output_func (#54914)

heitorschueroff committed 3 years ago

8b8c4096

Newer Older