strisunshinewentingwang/pytorch

Pull Requests Commits

[FSDP] Fix input grad propagation when using param mixed precision (#90921)

awgu committed 2 years ago

d04e3c99

Revert "[quant][pt2e] Add early prototype top level quantize_pt2e APIs (#90802)"

pytorchmergebot committed 2 years ago

9c912c7d

[inductor] Use --continue_on_fail when installing torchbench (#90922)

desertfire committed 2 years ago

fdc97330

[Dynamo][TIMM][Benchmarks] Fix TIMM `0.8.0dev` breaking the `timm_models.py` script's data config (#90404)

eqy committed 2 years ago

57e2090e

If a torch.* returns non-Tensor, make this unimplemented rather than assert. (#89918)

ezyang committed 2 years ago

e686a442

[quant][pt2e] Add early prototype top level quantize_pt2e APIs (#90802)

jerryzh168 committed 2 years ago

a66af1fe

Hack get_nbytes() to return 0 for sparse tensors as workaround for functionalization (#90702)

jbschlosser committed 2 years ago

201c36d8

Error messages for kernel selection (#90783)

Michael Gschwind committed 2 years ago

15c9df77

[ao][fx] fixing public v private qconfig_mapping_utils.py (#88399)

HDCharles committed 2 years ago

173accd1

Revert "Revert "[functorch] Refactor life handle storage (#90317)"" (#90856)

zou3519 committed 2 years ago

abc54f93

[inductor] Prevent blowup in inner_fn_str and extract_read_writes (#88933)

peterbell10 committed 2 years ago

81f351ac

[FSDP] Enable mixed hybrid/non-hybrid sharding strategies (#90846)

awgu committed 2 years ago

c4718e9b

[FSDP][Easy] Use `run_subtests` for hybrid shard test (#90859)

awgu committed 2 years ago

2f8c0cb2

replicate state_dict tests (#90868)

rohan-varma committed 2 years ago

b92975a6

[inductor] add conv_transpose2d unary fusion for cpu in inference mode (#90265)

chunyuan-w committed 2 years ago

d6fe9838

add conv_transpose2d pointwise(unary) fusion kernel (#90264)

chunyuan-w committed 2 years ago

85698d0a

[Quant] onednn backend switch to ideep new api without affacting performance (#90354)

Xia-Weiwen committed 2 years ago

9b89ff09

[CUDA 12] Fix the endif guard position for cusparse const descriptors (#90897)

xwang233 committed 2 years ago

79009cbc

[Composable API] `replicate`: cleanup _ddp.py (#90257)

Charlie Yan committed 2 years ago

98799ca0

Deeply rework WeakIdKeyDictionary (#90825)

ezyang committed 2 years ago

0b22f5ae

Don't put tracing state on Tensor (#90628)

ezyang committed 2 years ago

54563e62

inductor: sort the reads buf by name (#89744)

chunyuan-w committed 2 years ago

103029e0

fix cudnn RNN reproducibility problem (#90522)

cyyever committed 2 years ago

9fe050f3

fix a memory leak on return without free (#90372)

cyyever committed 2 years ago

dcfe7ff7

Reland Fix issue 38095 TODO in test_multiprocessing.py (#90741)

kit1980 committed 2 years ago

0ac0af02

Fix unittest rerun logic when checking for skipped tests (#90888)

huydhn committed 2 years ago

4e645516

Inductor cpp wrapper: change inputs args from tuple to vector (#90754)

chunyuan-w committed 2 years ago

2ba5c1d7

[FSDP][Easy] ufmt files (#90858)

awgu committed 2 years ago

39d9dd13

[CUDA] Use accumulate type to improve accuracy of grid_sample on half precision inputs (#90427)

xwang233 committed 2 years ago

670efb97

[cuDNN][cuDNN V8 API] (re-re-open) cuDNN V8 API on by default (#89022)

eqy committed 2 years ago

eecd621f

Newer Older