mirror of
https://github.com/zebrajr/pytorch.git
synced 2025-12-06 00:20:18 +01:00
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/49251 Since all ops are c10-full and use templated unboxing now, we don't need to codegenerate any unboxing logic anymore. Since this codegen was the only code using setManuallyBoxedKernel, we can also remove that functionality from KernelFunction, OperatorEntry and Dispatcher. ghstack-source-id: 119450486 Test Plan: waitforsandcastle Reviewed By: ezyang Differential Revision: D25502865 fbshipit-source-id: 49d009df159fda4be41bd02457d4427e6e638c10 |
||
|---|---|---|
| .. | ||
| templates | ||
| __init__.py | ||