strisunshinewentingwang/pytorch

Pull Requests Commits

[PyTorch] Enable template build at aten operator level (#53801)

cccclai committed 3 years ago

b7b481bd

ns for fx: add weight matching for linear fp16 emulation (#54257)

vkuzo committed 3 years ago

0a182119

ns for fx: add partial support for subgraphs with base_op_node (#54254)

vkuzo committed 3 years ago

182d8c37

ns for fx: create subgraph type (#54253)

vkuzo committed 3 years ago

454832e5

ns for fx: move shadow lstm test to new API (#53828)

vkuzo committed 3 years ago

9e8e744e

ns for fx: move shadow activations linear test to new API (#53819)

vkuzo committed 3 years ago

cfe73648

ns for fx: move shadow activations conv test to new API (#53818)

vkuzo committed 3 years ago

3dc8ba27

ns for fx: add support for lstm activation matching (#53779)

vkuzo committed 3 years ago

52a8075f

[FX] Normalize Python operators to `torch.` ops when called with Tensors (#54236)

James Reed committed 3 years ago

c656a5be

fx quant: fix bug with fusion patterns and disabling quantization (#54654)

vkuzo committed 3 years ago

b81e10a2

[FX] Garbage collect values in Interpreter (#54726)

James Reed committed 3 years ago

a28c7db9

Pin autocanceling GHA repo to specific commit (#54738)

janeyx99 committed 3 years ago

fd58ecec

[fuser] Support bfloat16 (#54571)

bertmaher committed 3 years ago

9db48021

[DDP logging] Prefer use of c10::Join (#54649)

rohan-varma committed 3 years ago

6b7652e2

lu_backward: more numerically stable and with complex support. (#53994)

nikitaved committed 3 years ago

dfc7fa03

Automated submodule update: tensorpipe (#54686)

facebook-github-bot committed 3 years ago

3bb0f1f3

[CMake] Simplify CPU architecture detection logic (#54637)

malfet committed 3 years ago

68bdeef2

[package] rename PackageExporter.external to PacakgeExporter.extern_modules (#54601)

suo committed 3 years ago

911b8b1b

Fix broken javadoc URL in README (#54434)

nairbv committed 3 years ago

9c60fc9c

Add doc warnings for default SELU gain (#54057)

mrTsjolder committed 3 years ago

a7c7fc96

Simplify creation of unary structured kernels. (#54592)

hameerabbasi committed 3 years ago

f1edaabc

Add GHA to cancel redundant GHA workflows except on master (#54689)

janeyx99 committed 3 years ago

71b9f2dd

Remove hacky wrapper for about 100 kernels (#54367)

wenleix committed 3 years ago

53596cdb

Handle stride > 1 with im2col in CUDA thnn conv2d (#54080)

t-vi committed 3 years ago

d12118c0

Revert D27327999: [pytorch][PR] Cancel redundant GHA workflows

samestep committed 3 years ago

0b0a5dd3

Cancel redundant GHA workflows (#54685)

samestep committed 3 years ago

f251bb40

[Gradient Compression] Add logging for gradient compression stats. (#54647)

Mark Astley committed 3 years ago

4bf90558

Ensure torch.futures.wait_all exits early on error. (#53953)

pritamdamania committed 3 years ago

267fc27d

Make SharedCache thread-safe (#53750)

gmyr committed 3 years ago

93bbbecc

Introduce a fluent API to construct tensors from external data. (#54530)

cbalioglu committed 3 years ago

9029d0d7

Older