strisunshinewentingwang/pytorch

Pull Requests Commits

TorchDynamo: weight prepack for onednn convolution external call (#88988)

XiaobingSuper committed 1 year ago

07151a6b

Revert "Dont clone unmutated args in triton autotuning (#89519)" (#89652)

ezyang committed 1 year ago

0884fdab

Reenable fake_tensor_propagation on test_cudnn_rnn (#89644)

ezyang committed 1 year ago

4a16f8cd

Run optimizer tests with fake tensors (#89643)

ezyang committed 1 year ago

fc7dcb68

Force test_rng_state to run with fake tensor prop (#89641)

ezyang committed 1 year ago

9b13508e

Easy: These tests work with fake_tensor_propagation on (#89640)

ezyang committed 1 year ago

c6be06d9

Support unspecialized integers with dynamic shapes (#89639)

ezyang committed 1 year ago

6fb6eb0a

Cond capture with fake tensors actually works; don't raise in this case (#89638)

ezyang committed 1 year ago

0c96841a

[test_nn] split pruning tests from test_nn (#89590)

kshitij12345 committed 1 year ago

d3c012f4

Added vectorized CPU code for uint8_t datatype. (#89284)

alexsamardzic committed 1 year ago

83666f16

Update SyncBatchNorm _all_gather_base to all_gather_into_tensor (#89521)

H-Huang committed 1 year ago

94975527

Remove fake_tensors_available (#89637)

ezyang committed 1 year ago

94a88b53

Fix segfault when swapping custom allocator (#89613)

Emilio Castillo committed 1 year ago

1c8b0779

Make pytest work again on test/dynamo (#89631)

ezyang committed 1 year ago

fd279fe8

Mention discrepency between original impl and our impl of RAdam (#89575)

albanD committed 1 year ago

c3e85d87

Suppress guards on as_strided call only. (#89569)

ezyang committed 1 year ago

860bae49

Added log1p for complex in c10 (#89214)

mfkasim1 committed 1 year ago

1588ea0d

[LTC] Refine MetricsArena::Reset (#89608)

alanwaketan committed 1 year ago

4f5c4c02

Upgrade nightly wheels to ROCm5.3 (#89101)

jithunnair-amd committed 1 year ago

a8629a1c

Use fx.replace_pattern for removing empty_like+fill in nvFuser+PrimTorch execution (#89132)

IvanYashchuk committed 1 year ago

c0d81aa7

[QAT] Check the value of numel to avoid segfault (#81547)

hguandl committed 1 year ago

b515c1d9

quantization: deprecate observer compute_dtype and replace with is_dynamic (#85431)

vkuzo committed 1 year ago

22a1b5e2

[Dynamo] Fix bug of using customized torch.autograd.Function (#89397)

yanboliang committed 1 year ago

e4ccec6e

Disable optimizer tracing, enable for tests only (#89500)

mlazos committed 1 year ago

903ae457

Expose to python the backward AD view_func (#89586)

albanD committed 1 year ago

c79489c8

Symintify `embedding` (#89327)

nkaretnikov committed 1 year ago

4cb6bbbe

nnc: fix Store if value is fp32 while buf is bf16 (#86788)

chunyuan-w committed 1 year ago

9c867eae

Symintified layer_norm (#89466)

tissue3 committed 1 year ago

f0e5bc4b

Install missing VSX headers (POWER) (#85547)

Flamefire committed 1 year ago

fdb2dd11

[ONNX] Move two headers from .h to .cc (#86852)

wschin committed 1 year ago

e922bd4e

Newer Older