strisunshinewentingwang/pytorch

Pull Requests Commits

[BE] Add regression test for aten shared build (#96697)

malfet committed 1 year ago

5a8a4030

Revert "Enable thp(transparent huge pages) for buffer sizes >=2MB (#95963)"

pytorchmergebot committed 1 year ago

a22b92d8

Update Exceptions.cpp (#96031)

co63oc committed 1 year ago

86a9fe8a

Don't specialize torch.Size with specialize_int = False (#96419)

ezyang committed 1 year ago

507feb80

Return Live Data Pointers from Checkpoint, swap onto tensors (#95020)

eellison committed 1 year ago

da265652

Handle additional live allocations not in checkpointed state (#94943)

eellison committed 1 year ago

1cc32aed

Checkpoint CUDA Allocator Private Pool State (#94653)

eellison committed 1 year ago

d798de2b

[MPS] Fix flip where no dims need to be flipped (#96605)

qqaatw committed 1 year ago

c95bcb66

[Quant][fx] Remove unused is_qat args in prepare_fx (#96631)

andrewor14 committed 1 year ago

ca7e5332

[CUDA][cuFFT] Minor fix for cuFFT plan cache docs (#96373)

eqy committed 1 year ago

6e3e22d5

Update conv_fused.py (#95551)

co63oc committed 1 year ago

6a492908

Make linter image available on ECR when rebuilding (#96671)

huydhn committed 1 year ago

ae4d6909

Add torch.nn.LayerNorm() to documented list of supported nested tensor ops (#96434)

drisspg committed 1 year ago

f330281f

[MPS] Add C++ API support for MPS backend (#96668)

razarmehr committed 1 year ago

069ace13

Update CUDAGraph.cpp (#96029)

co63oc committed 1 year ago

c28b224e

Fix the typo for the docstring of args in the observer (#95887)

yiliu30 committed 1 year ago

2ea0cb12

Gramatically updated the tech docs (#92896)

SridharPerepu committed 1 year ago

9159599c

[dynamo][dashboard] Clear local changes before pulling git repos (#96667)

davidberard98 committed 1 year ago

1d792288

Incorrect links #96515 (#96536)

RishabhSpark committed 1 year ago

16a16d18

test_foreach: remove `skipMeta` (#96599)

crcrpar committed 1 year ago

a48d518e

[FSDP][optim_state_dict] Make FSDP optim_state_dict aware of DDP prefix (#96415)

fegin committed 1 year ago

f5a0b31a

[pytorch][coreml] Use from_blob instead of empty in pack_outputs (#96564)

kylesyoon committed 1 year ago

b9921994

[CUDA12] Autograd engine use current device only (#92354)

Aidyn-A committed 1 year ago

c69b3b8d

Changed flop formulas for flop counter to take in shapes directly (#96565)

Chillee committed 1 year ago

31137a63

[inductor] fixes addmm pattern matcher to exclude non-conformant patt… (#96634)

ngimel committed 1 year ago

3f1efade

Support randn_like() for NT (#96528)

jbschlosser committed 1 year ago

30d56dd8

Add a new knob to separately enable the autotuning in Triton. (#96440)

nmacchioni committed 1 year ago

f673ad6d

Add triton to relevant packages (#96663)

ezyang committed 1 year ago

4454655a

Convenience script for getting correct Triton nightly binary (#96669)

ezyang committed 1 year ago

a8d1eb19

Revert all_reduce workaround as it might be causing issues on other parts of the codebase (#96460)

Rodrigo Kumpera committed 1 year ago

120c6f66

Newer Older