strisunshinewentingwang/pytorch

Pull Requests Commits

Revert "Flash Attention v2 (#105602)" (#108827)

huydhn committed 1 year ago

a9c663c2

Improve torch.cuda.amp type hints (#108630)

ringohoffman committed 1 year ago

e40d6ae0

Back out "Horizontally fuse input concatenation (#108115)" (#108793)

mlazos committed 1 year ago

6c726040

Revert "[inductor] Switch to use the runtime interface for AOTInductor testing (#108663)"

pytorchmergebot committed 1 year ago

428f5f9e

[dynamo] Move global state guards to C++ (#108624)

jansel committed 1 year ago

4965fffe

[vision hash update] update the pinned vision hash (#108818)

pytorchupdatebot committed 1 year ago

258bc2d8

[AOTInductor] Include constants in AOTInductor .so file. (#108473)

muchulee8 committed 1 year ago

30a33b76

Revert "[dynamo][finishing colesbury's PR 100642] Guard on nn.Module dicts and type (#108528)"

pytorchmergebot committed 1 year ago

72f24d00

Revert "Revert "Flash Attention v2 (#105602)" (#108827)"

pytorchmergebot committed 1 year ago

e45b2901

Revert "Flash Attention v2 (#105602)" (#108827)

huydhn committed 1 year ago

24e9bbe2

fixed nn.Module.to type hint (#108767)

randolf-scholz committed 1 year ago

8391e3fb

[Dynamo][Test]Add a testcase for module with training state (#108750)

youkaichao committed 1 year ago

f90444cf

[dynamo] Add "Torch-Compiled Region" profiler event (#108462)

davidberard98 committed 1 year ago

dec2b267

Revert "[dynamo] Add BACKEND_MATCH guard to detect and recompile when backend changes (#107337)"

pytorchmergebot committed 1 year ago

38fcf77a

fix returning in void function (#108774)

cyyever committed 1 year ago

e3280a7c

[C10d] Fix TCPSTore::wait to be robust to interruptions. (#108425)

Rodrigo Kumpera committed 1 year ago

a6dab862

[Lint] Auto format graph_module.py (#108594)

jingsh committed 1 year ago

fc2b9800

Decompose/add reference for `view_as_complex` (#108005)

Fidget-Spinner committed 1 year ago

c458fa0d

[inductor] Switch to use the runtime interface for AOTInductor testing (#108663)

desertfire committed 1 year ago

366ce589

enforce equalities (#108429)

avikchaudhuri committed 1 year ago

c55cb29b

Run mm decomposition tests for CPU and GPU (#108620)

mikekgfb committed 1 year ago

247c603d

[dynamo] Add BACKEND_MATCH guard to detect and recompile when backend changes (#107337)

ydwu4 committed 1 year ago

1a64ec7d

[c10d] Add TCPSTore libuv backend support to c10d rendezvous. (#108284)

Rodrigo Kumpera committed 1 year ago

b26af5d5

[Dev Container][CUDA]Fix linker path (#108766)

youkaichao committed 1 year ago

96d269ea

Add better error messaging to scaled_mm (#108454)

drisspg committed 1 year ago

09a17c51

fall back to eager on `NotImplementedError` (#107863)

ev-br committed 1 year ago

1f205319

Revert "[inductor] Add ir.Scan and lower aten.cumsum on CUDA (#106581)"

pytorchmergebot committed 1 year ago

8ba23e48

Fix expected test failures for predispatch export nested cond and out_dtype (#108715)

ydwu4 committed 1 year ago

774c8229

[inductor] Add ir.Scan and lower aten.cumsum on CUDA (#106581)

peterbell10 committed 1 year ago

53a27021

Remove fixed skips (#108674)

eellison committed 1 year ago

ab9fb03d

Newer Older