strisunshinewentingwang/pytorch

Pull Requests Commits

export helper funcs in foreach ops (#102928)

heidongxianhua committed 1 year ago

978a2f2b

[PT2 Dynamo Hackathon] Fix simple bug in inline dict (#103187)

tugsbayasgalan committed 1 year ago

91e82ba0

add github action to upload alerts to rockset / aws (#102995)

PaliC committed 1 year ago

49450fe0

Make index_add_ error if input source shape is wrong (#100321)

tfsingh committed 1 year ago

d2d03f0f

Enable torch.nn.init._calculate_correct_fan in dynamo tracing (#103182)

fduwjj committed 1 year ago

52e310f7

GHA setup-linux should always be pair with teardown-linux (#103216)

huydhn committed 1 year ago

676210a1

Two small mem_eff bug fixes (#103201)

drisspg committed 1 year ago

2868a5d0

[quant][pt2] Add prepare QAT test for resnet18 (#103020)

andrewor14 committed 1 year ago

9508e60c

fix amp in inference in benchmarking suite (#103220)

eellison committed 1 year ago

18e4a466

[dtensor] fix allgather unpadding logic (#103219)

wanchaol committed 1 year ago

8585784a

[FSDP]Remove dim_group from device_mesh init (#103218)

wz337 committed 1 year ago

d5142c52

[data_loader] Extra signal handlers in DataLoader.cpp should be added on top rather than replacing defaults (#103164)

andriigrynenko committed 1 year ago

6acb8d3d

Make HigherOrderOperator stop appearing like torch.ops.* in FX (#103108)

zou3519 committed 1 year ago

194262ee

Add selector.is_et_kernel_key_selected (#103184)

kirklandsign committed 1 year ago

47cfcf56

[PyTorch][HAM]: Deprecate functionalize (#103053)

XuanQi committed 1 year ago

c37f02f6

[inductor][easy] raise register spill threshold (#103190)

shunting314 committed 1 year ago

0900782f

[inductor] Fix correctness issues with pre_grad and context managers (#103051)

jansel committed 1 year ago

8c5d97d3

[DTensor] Allow DTensor support cuda-like device (#102468)

shaoyf42 committed 1 year ago

17737f9d

Fix Graph Break on builtin comparison on NNModule (#103176)

msaroufim committed 1 year ago

790f5732

Pretty dataclass dynamo explain (#102869)

msaroufim committed 1 year ago

95fced44

Small operatorbench improvements (#103110)

eellison committed 1 year ago

2baadc2a

[ROCm] force HIP context initialization for inductor UTs (#103149)

jataylo committed 1 year ago

e936277c

Use gcc9 in linux-bionic-cuda12_1-py3_10-gcc9-build workflows (#103075)

atalman committed 1 year ago

376cf796

Enable torch.get_autocast_gpu_dtype in Dynamo tracing (#103166)

fduwjj committed 1 year ago

c454534d

Enable torch.is_complex in Dynamo tracing (#103154)

fduwjj committed 1 year ago

b5021ba9

[quant][pt2] Add test for inplace add (#102867)

andrewor14 committed 1 year ago

2e8d2a2e

Fix outdated log settings in doc (#102285) (#102286)

Weiming Zhao committed 1 year ago

28f43c76

[PT2][Quant] Use composble quantizer for embedding + static conv + dynamic (#103116)

kimishpatel committed 1 year ago

471407cf

[MPS] Prerequisite for MPS C++ extension (#102483)

qqaatw committed 1 year ago

3c0072e7

[dashboard] Bring back inference perf measurement as nightly (#103151)

desertfire committed 1 year ago

0c9117a6

Newer Older