strisunshinewentingwang/pytorch

Pull Requests Commits

ezyang committed 2 years ago

696f8b85

single process repro

ezyang committed 2 years ago

6e73c21d

Asserts and stuff

ezyang committed 2 years ago

b482c60b

ezyang committed 2 years ago

475b0471

voznesenskym committed 2 years ago

d59464a9

Unit test fixes

voznesenskym committed 2 years ago

3610c18f

[UPDATED PROTOTYPE] Use dynamo fake tensor mode in aot_autograd, move aot_autograd compilation to lowering time

voznesenskym committed 2 years ago

ee1012ac

Use isinstance test rather than exact type test for wrap to fake

voznesenskym committed 2 years ago

11629b17

Make aot_module_simplified accept fake tensors

voznesenskym committed 2 years ago

163cbaa4

Change aot_module_simplified to take take arguments directly

voznesenskym committed 2 years ago

eb2cc384

Don't suppress exceptions from backends

voznesenskym committed 2 years ago

9aa3dc8b

Simplify aot_module_simplified by removing top_args/top_kwargs

ezyang committed 2 years ago

83781af9

Don't support kwargs at runtime in aot_module_simplified

ezyang committed 2 years ago

4b9d79c0

Delay verify correctness wrapping to call site.

ezyang committed 2 years ago

33ece499

Remove fake_tensor_propagation (#89646)

ezyang committed 2 years ago

69043247

xfail maml test, instead of running it without fake tensor prop (#89645)

ezyang committed 2 years ago

1aa1014b

[vision hash update] update the pinned vision hash (#89667)

pytorchmergebot committed 2 years ago

a048913e

TorchDynamo: weight prepack for single conv (#89209)

XiaobingSuper committed 2 years ago

3b3ebcd0

TorchDynamo: weight prepack for mkl linear (#89109)

XiaobingSuper committed 2 years ago

0c4f3db7

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

XiaobingSuper committed 2 years ago

07151a6b

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

ezyang committed 2 years ago

0884fdab

Reenable fake_tensor_propagation on test_cudnn_rnn (#89644)

ezyang committed 2 years ago

4a16f8cd

Run optimizer tests with fake tensors (#89643)

ezyang committed 2 years ago

fc7dcb68

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

ezyang committed 2 years ago

9b13508e

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

ezyang committed 2 years ago

c6be06d9

Support unspecialized integers with dynamic shapes (#89639)

ezyang committed 2 years ago

6fb6eb0a

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

ezyang committed 2 years ago

0c96841a

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

kshitij12345 committed 2 years ago

d3c012f4

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

alexsamardzic committed 2 years ago

83666f16

Update SyncBatchNorm _all_gather_base to all_gather_into_tensor (#89521)

H-Huang committed 2 years ago

94975527

Older