strisunshinewentingwang/pytorch

Pull Requests Commits

Automated submodule update: FBGEMM (#54486)

facebook-github-bot committed 3 years ago

a4a21e7d

[jit][tensorexpr] Added aten::batch_norm into fuser when in inference mode (#54204)

huiguoo committed 3 years ago

2a538971

[AutoAccept][Codemod][FBSourceClangFormatLinter] Daily `arc lint --take CLANGFORMAT`

generatedunixname89002005325676 committed 3 years ago

fee470d8

Enabled BFloat16 support for argmax & argmin on both CPU & CUDA (#52582)

sanchit committed 3 years ago

f2a38a0e

Fix onnx warning message (#54371)

Peiyuan Liao committed 3 years ago

3519625a

Grammatically update tech docs (#54370)

Ioana Tivadar committed 3 years ago

1041fdd0

[PyTorch] Update Bazel build for TensorPipe (#54416)

lw committed 3 years ago

8518b0ee

[Gradient Compression] Remove cuda.syncrhonize in batched powerSGD (#54482)

Yi Wang committed 3 years ago

c22fc448

Delete all unnecessary singular Math entries (#54436)

ezyang committed 3 years ago

6d002719

s/StructuredNativeFunctions/NativeFunctionsGroup/ (#54427)

ezyang committed 3 years ago

6e8c4ad7

Rejigger to use NativeFunctionsGroup even without structured: True (#54426)

ezyang committed 3 years ago

bf2ca35f

Move compute_native_function_declaration to its own dest module (#54419)

ezyang committed 3 years ago

c00d66f7

Replace some tensor.device().is_cpu() calls with direct tensor.is_cpu() (#54397)

ezyang committed 3 years ago

349a17f1

[5/n][torch/elastic][upstream] Move torchelastic/agent to torch/distributed/elastic/agent (#54343)

aivanou committed 3 years ago

77ccd4f9

Port index_copy from TH to ATen (#52203)

lezcano committed 3 years ago

58703461

[Static Runtime] Fix bug in reshape_copy (#54467)

Hao Lu committed 3 years ago

52abd3bd

Only allow hub.load() from original repo. (#54451)

Ailing Zhang committed 3 years ago

c411017a

[JIT] Add Reinplacing to MKLDNN Subgraphs (#53908)

Elias Ellison committed 3 years ago

9be4c75f

use reshape when possible in broadcasting (#53326)

Elias Ellison committed 3 years ago

81c6e5fb

Avoid dispatch overhead in call to mkldnn convolution (#52614)

Elias Ellison committed 3 years ago

18c04a3f

[PyTorch][JIT] Less shared_ptr use in dictConstruct (#54110)

swolchok committed 3 years ago

3959d393

[PyTorch] Avoid extra intrusive_ptr copy in IValue::toIntrusivePtr (#54124)

swolchok committed 3 years ago

7e33dc34

Automated submodule update: FBGEMM (#54447)

facebook-github-bot committed 3 years ago

568d43b9

Get rid of {Cpu,Cuda}{Channel,Context} in tensorpipe_agent. (#54432)

Lucas Hosseini committed 3 years ago

decbdf7b

Export torch::jit::toIValue (#54449)

Christian Puhrsch committed 3 years ago

2668149b

fix comparison of narrow type with wide type in loop condition (#53951)

frank-dong-ms committed 3 years ago

92770d25

Migrate kernels with Tensor? to C10 full dispatcher (#54263)

wenleix committed 3 years ago

edfc787d

Add device_of overload for optional<Tensor> (#54262)

wenleix committed 3 years ago

5a271991

Use c10::ArrayRef instead of std::vector for the jit::unpickle's tensor_table. (#54428)

cbalioglu committed 3 years ago

2130f4cc

Delete dead TensorOptions::key_set (#54004)

ezyang committed 3 years ago

4919fecf

Newer Older