strisunshinewentingwang/pytorch

Pull Requests Commits

[cuDNN][cuDNN V8 API] Fix `benchmark_limit` ignoring failed kernels in FIND (#91032)

eqy committed 1 year ago

ce4900f3

Vectorize exmp1 and log1p (#91074)

EikanWang committed 1 year ago

856651dd

[Dynamo] Support torch.get_default_dtype (#89790)

yanboliang committed 1 year ago

490c1cf6

Re-enable optimizers (#90709)

mlazos committed 1 year ago

1accd915

[Quant][FX] Lower QLinearLeakyReLU for onednn backend (#88668)

Xia-Weiwen committed 1 year ago

9ca41a98

Fix CSR with int32 indices to CSC conversion (#91061)

pearu committed 1 year ago

8004f934

[Checkpoint][Test] Add 2d DCP model state checkpoint test (save/load) (#91046)

wz337 committed 1 year ago

6be1e433

Introduce guardexpr, aot autograd guarding of duplicates into torch._guards (#90955)

voznesenskym committed 1 year ago

b72caf31

Add dynamic shapes benchmark accuracy to CI (#90444)

ezyang committed 1 year ago

212873c6

[Composable API] Enable composable `fully_shard` submodules in `replicate` parent module (#90711)

yhcharles committed 1 year ago

a1a2f548

[Checkpoint][nit] Fix test_fsdp_optim_state.py test name (#90943)

wz337 committed 1 year ago

3229713c

Revert "Add dynamic shapes benchmark accuracy to CI (#90444)"

pytorchmergebot committed 1 year ago

e2377c83

Add dynamic shapes benchmark accuracy to CI (#90444)

ezyang committed 1 year ago

85db031e

[reland3][dynamo] Revert "Revert "[reland][dynamo] use optimizers correctly in benchmar… (#90956)

mlazos committed 1 year ago

7c524221

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

pytorchmergebot committed 1 year ago

78efde92

[Quant][FX] Add backend config for onednn backend and fuse Linear-LeakyReLU (#88665)

Xia-Weiwen committed 1 year ago

7b0ec67e

[Checkpoint] Fix checkpoint test test_fsdp_optim_state.py (#91036)

wz337 committed 1 year ago

bfa223aa

Remove duplicate line (#91006)

ZainRizvi committed 1 year ago

1d948787

[reland][quant][pt2e] Add early prototype top level quantize_pt2e APIs (#91035)

jerryzh168 committed 1 year ago

f7b384cc

fix default partitioner: save sizes instead of tensor for backward when possible (#91012)

bdhirsh committed 1 year ago

4ab81ae8

Save and restore tracked_fakes (#90995)

ezyang committed 1 year ago

1609b954

[functorch] add composition-of-3-transform tests for autograd_function (#90962)

zou3519 committed 1 year ago

ed589dd8

Fix comment about get_fw_grad_mode() only being used in custom Function (#90790)

zou3519 committed 1 year ago

e1c799ff

Add VmapInterpreter.randomness (in pyfunctorch) provide it in info object (#90789)

zou3519 committed 1 year ago

ffa37c9f

set -Winconsistent-missing-override for builds (#89851)

mikey dagitses committed 1 year ago

8bd959e4

lint transformer.py (#91048)

davidberard98 committed 1 year ago

93cb5806

[dynamo] turn torch.backends.cudnn.is_acceptable into a constant (#90323)

davidberard98 committed 1 year ago

5d70d128

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

pytorchmergebot committed 1 year ago

7d3f2b79

Allow Process Group to support multiple backends (#88330) (#90997)

H-Huang committed 1 year ago

7a0f29b7

[dynamo] Refactor how autocast parameters are binded (#90953)

desertfire committed 1 year ago

93ac8c4a

Newer Older