Retire attention-mask custom operator - #21802
Conversation
Summary: Retire the Jarvis attention-mask custom operator and all confirmed dead scaffolding. Remove both the ExecuTorch `jarvis_nn_ops::attention_mask.out` runtime path and the unused eager `jarvis_nn_ops::attention_mask` wrapper. Delete their generic and HiFi kernels, tests, build targets, registrations, dependencies, and documentation. Remove custom-op-only PYE libraries, the orphan ASR encoder runtime package, unused ASR aggregate/HTP targets, and downstream references while preserving the live Jarvis NN aggregates and lean/non-HTP ASR operator bundles. Differential Revision: D115805406
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21802
Note: Links to docs will display an error until the docs builds have been completed. ❌ 2 New FailuresAs of commit cde4058 with merge base 43f89fb ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
@mcremon-meta has exported this pull request. If you are a Meta employee, you can view the originating Diff in D115805406. |
This PR needs a
|
Summary:
Retire the Jarvis attention-mask custom operator and all confirmed dead scaffolding.
Remove both the ExecuTorch
jarvis_nn_ops::attention_mask.outruntime path and the unused eagerjarvis_nn_ops::attention_maskwrapper. Delete their generic and HiFi kernels, tests, build targets, registrations, dependencies, and documentation. Remove custom-op-only PYE libraries, the orphan ASR encoder runtime package, unused ASR aggregate/HTP targets, and downstream references while preserving the live Jarvis NN aggregates and lean/non-HTP ASR operator bundles.Differential Revision: D115805406