Yidi Wu c27882ffa8 [torchbind] always fakify script object by default in non-strict export (#127116) ... This diff can be risky for internal tests: any torchbind class that hasn't registered a fake class will fail and we should fix them. We've gained some confidence that this can work e2e by implementing FakeTensorQueue for TBE models in sigmoid with [D54210823](https://www.internalfb.com/diff/D54210823). Differential Revision: [D57991002](https://our.internmc.facebook.com/intern/diff/D57991002) Pull Request resolved: https://github.com/pytorch/pytorch/pull/127116 Approved by: https://github.com/zou3519 ghstack dependencies: #127113, #127114		2024-06-03 21:38:57 +00:00
..
db
pass_infra
passes	Copy the constant folding pass to the pass under export/passes folder (#127456)	2024-05-30 18:04:08 +00:00
serde	[export] Handle serializing duplicate getitem nodes (#127633)	2024-06-03 17:25:51 +00:00
__init__.py	[export] Disable backend decomps for capture_pre_autograd (#127120)	2024-05-28 16:37:13 +00:00
converter.py	[ts migration] support aten::dim, aten::len, aten::__getitem__ (#127593)	2024-06-02 00:36:33 +00:00
error.py
exported_program.py
non_strict_utils.py	[torchbind] always fakify script object by default in non-strict export (#127116)	2024-06-03 21:38:57 +00:00
pass_base.py
tools.py
utils.py	[Fix]: populate input parameter name when convert TorchScript to ExportedProgram (#126787)	2024-05-28 17:33:44 +00:00
verifier.py	Add a registry for GraphModuleSerializer (#126550)	2024-05-29 03:12:48 +00:00
wrappers.py