Use new API to register custom ops for llama model (#2840)
Summary: Using the following 2 APIs: * `EXECUTORCH_LIBRARY` to replace the need of a yaml file. With this macro we can directly register a custom kernel into ExecuTorch runtime. * `WRAP_TO_ATEN` allows custom op authors to use the same kernel for ExecuTorch and PyTorch. This can be helpful during debugging. Pull Request resolved: https://github.com/pytorch/executorch/pull/2840 Test Plan: Rely on the new CI job `test_llama` with `xnnpack+kv+custom` option. Reviewed By: kimishpatel Differential Revision: D55713944 Pulled By: larryliu0820 fbshipit-source-id: eff68926c16a1b2a39f36c73eab1da4e983a25c6
M
Mengwei Liu committed
020d8bee8fddf837006c4e35bf7dff5278df2e24
Parent: 4d9863f
Committed by Facebook GitHub Bot <facebook-github-bot@users.noreply.github.com>
on 4/6/2024, 4:26:49 AM