Repository navigation
Lower tanh with the existing unary kernel - #24452
JakeStevens wants to merge 1 commit into
Conversation
Summary: ## Ulterior Motive Unblock the tanh activation in native MaxViT execution. ## Rationale **What**: Dispatch `aten.tanh.default` through the existing unary lowering and `ggml_tanh` kernel. **Why**: The graph builder rejects the operator even though ggml already implements it. ## Details - Reuse the existing floating-point validation and unary input handling. - Check contiguous and transposed inputs with cached and uncached execution. - Cover signed zero, infinities, NaN and integer-input rejection. Authored with AI using Metamate on agentcloud. Differential Revision: D129100402
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/24452
Note: Links to docs will display an error until the docs builds have been completed. ❌ 2 New Failures, 2 Unclassified FailuresAs of commit 8d4df08 with merge base 822cb60 ( NEW FAILURES - The following jobs have failed:
UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
@JakeStevens has exported this pull request. If you are a Meta employee, you can view the originating Diff in D129100402. |
This PR needs a
|
Summary:
Ulterior Motive
Unblock the tanh activation in native MaxViT execution.
Rationale
What: Dispatch
aten.tanh.defaultthrough the existing unary lowering andggml_tanhkernel.Why: The graph builder rejects the operator even though ggml already implements it.
Details
Authored with AI using Metamate on agentcloud.
Differential Revision: D129100402