strisunshinewentingwang/pytorch

Pull Requests Commits

set workspace size to 4M (#74159)

jianyuh committed 2 years ago

50ae5c91

Remove expired c10d::broadcast backward compatibility check (#84107)

mrshenli committed 2 years ago

87738f20

move internal only PyTorch test defs into fb/ subdirectories (#84605)

mikey dagitses committed 2 years ago

99b7eb4d

Add xla/ folder to gitignore (#84632)

lezcano committed 2 years ago

d3d163af

Call jit decomposition in VariableType to increase forward AD coverage (#84151)

soulitzer committed 2 years ago

42d99e6f

[reland] Move decompositions and helpers for jvp from functorch into core (#84581)

soulitzer committed 2 years ago

e31ad1c2

optimize householder product backward to be more memory-efficient (#84627)

nikitaved committed 2 years ago

3eb16509

[c10d] Fix docstring of scatter_object_list (#84596)

Rodrigo Kumpera committed 2 years ago

e96fb5d5

[composite compliance] fix linalg.eigvals (#84137)

zou3519 committed 2 years ago

a47bc96f

Add scatter_ to CommTensor (#84606)

mrshenli committed 2 years ago

89c4654b

Add broadcast_ to CommTensor (#84604)

mrshenli committed 2 years ago

f43c38bd

Add reduce_scatter_ to CommTensor (#84592)

mrshenli committed 2 years ago

a24d7a85

Supported nested lists in CommTensor and enable tracing allgather_ (#84585)

mrshenli committed 2 years ago

e4519548

Add mkl implementation for exponential on CPU (#69967)

CaoE committed 2 years ago

189768ed

Add alias info to torch._C (#84580)

sypneiwski committed 2 years ago

9e7af4e8

Detect `__code__` a bit more reliably. (#84610)

silvasean committed 2 years ago

ec3939a6

[composite compliance] linalg_householder_product (#84180)

kshitij12345 committed 2 years ago

07d398fb

[BE] Use `teardown-linux`/`chown` actions for binary builds (#84449)

malfet committed 2 years ago

045ebc77

nvfuser torchbench patch (#84411)

jjsjann123 committed 2 years ago

1a33e944

Add ShouldSyncTensor interface (#84418)

antoniojkim committed 2 years ago

7c3102f3

[c10d] Add a soft error handling mode (#84386)

kwen2501 committed 2 years ago

c4e0c927

Add deterministic impl of `scatter_add` CUDA for all input sizes (#79466)

kurtamohler committed 2 years ago

5b58140d

[vision hash update] update the pinned vision hash (#83900)

pytorchmergebot committed 2 years ago

039b0146

Throw on data dependent ops (#83567)

eellison committed 2 years ago

15c5baf8

[torchdynamo hash update] update the pinned torchdynamo hash (#84613)

pytorchmergebot committed 2 years ago

0be77d54

Make CommTensor Generic to arguments and outputs structures (#84576)

mrshenli committed 2 years ago

b168c4fa

[BE] Delete `Check for new workflow" check (#84608)

malfet committed 2 years ago

00e02280

Add watchdog to TorchElastic agent and trainers (#84081)

bchen2020 committed 2 years ago

06ebe2d5

ONNX: fix default function value in _optimize_graph (#83996)

ThibaultCastells committed 2 years ago

d9ceda49

[nnapi] remove unused field 'order_' in nnapi.h (#84067)

kirklandsign committed 2 years ago

16f8dc00

Newer Older