Add MLX op handler for aten.hardtanh

[Jah-yee]

Good day

This PR adds support for aten.hardtanh in the MLX delegate backend, addressing issue #18921.

Summary

Add a decomposed handler for aten.hardtanh that uses the existing ClipNode to implement bounded activation on Metal GPU.

Changes

• Add _hardtanh_handler in backends/mlx/ops.py using the decomposition: hardtanh(x, min_val, max_val) = clamp(x, min_val, max_val)
• Default values: min_val=-1, max_val=1
• Add test coverage for hardtanh operation

Testing

python -m executorch.backends.mlx.test.run_all_tests -k hardtanh

References

• Fixes: Good First Issue: Add MLX Op Handler for aten.hardtanh #18921

Thank you for your attention. If there are any issues or suggestions, please leave a comment and I will address them promptly.

Warmly,
RoomWithOutRoof

cc @metascroy

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/18986

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• Rolling out OSDC (ARC) runners on pull & trunk workflows in PyTorch main

❌ 19 New Failures, 2 Cancelled Jobs, 1 Pending

As of commit bed3054 with merge base 9ca0ff1 ():

NEW FAILURES - The following jobs have failed:

• MLX / test-mlx / test-mlx (gh)
  AttributeError: The underlying op of 'aten.bitwise_or' has no overload name 'default'
• MLX / test-mlx-llm (unsloth/gemma-3-1b-it, gemma3-1b, false, 4w) / test-mlx-llm-gemma3-1b-4w (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-llm (unsloth/gemma-3-1b-it, gemma3-1b, false, nvfp4) / test-mlx-llm-gemma3-1b-nvfp4 (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-llm (unsloth/gemma-3-1b-it, gemma3-1b, true, 4w) / test-mlx-llm-gemma3-1b-custom-4w (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-llm (unsloth/gemma-3-1b-it, gemma3-1b, true, nvfp4) / test-mlx-llm-gemma3-1b-custom-nvfp4 (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-llm (unsloth/Llama-3.2-1B-Instruct, llama-1b, false, 4w) / test-mlx-llm-llama-1b-4w (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-llm (unsloth/Llama-3.2-1B-Instruct, llama-1b, false, nvfp4) / test-mlx-llm-llama-1b-nvfp4 (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-llm (unsloth/Llama-3.2-1B-Instruct, llama-1b, true, 4w) / test-mlx-llm-llama-1b-custom-4w (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-llm (unsloth/Llama-3.2-1B-Instruct, llama-1b, true, nvfp4) / test-mlx-llm-llama-1b-custom-nvfp4 (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-llm (unsloth/Qwen3-0.6B, qwen3-0.6b, false, 4w) / test-mlx-llm-qwen3-0.6b-4w (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-llm (unsloth/Qwen3-0.6B, qwen3-0.6b, false, nvfp4) / test-mlx-llm-qwen3-0.6b-nvfp4 (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-llm (unsloth/Qwen3-0.6B, qwen3-0.6b, true, 4w) / test-mlx-llm-qwen3-0.6b-custom-4w (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-llm (unsloth/Qwen3-0.6B, qwen3-0.6b, true, nvfp4) / test-mlx-llm-qwen3-0.6b-custom-nvfp4 (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2
• MLX / test-mlx-parakeet / test-mlx-parakeet (gh)
  AttributeError: The underlying op of 'aten.bitwise_or' has no overload name 'default'
• MLX / test-mlx-qwen35-moe / test-mlx-qwen35-moe (gh)
  AttributeError: The underlying op of 'aten.bitwise_or' has no overload name 'default'
• MLX / test-mlx-stories110m / test-mlx-stories110m (gh)
  AttributeError: The underlying op of 'aten.bitwise_or' has no overload name 'default'
• MLX / test-mlx-voxtral / test-mlx-voxtral (gh)
  AttributeError: The underlying op of 'aten.bitwise_or' has no overload name 'default'
• MLX / test-mlx-voxtral-realtime / test-mlx-voxtral-realtime (gh)
  AttributeError: The underlying op of 'aten.bitwise_or' has no overload name 'default'
• MLX / test-mlx-whisper / test-mlx-whisper (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 2

CANCELLED JOBS - The following jobs were cancelled. Please retry:

• pull / unittest / windows / windows-job (gh)
• pull / unittest-editable / windows / windows-job (gh)

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[nil-is-all]

@claude review

[claude]

Claude finished @nil-is-all's task in 3m 26s —— View job

---

Review of PR #18986: Add MLX op handler for aten.hardtanh

• Read changed files and gather context
• Review existing op handler patterns in MLX backend
• Provide review feedback

---

Issue: Unrelated bitwise_or change included

The diff includes an unrelated addition at backends/mlx/ops.py:457:

([torch.ops.aten.bitwise_or.default], BitwiseOrNode, "aten.bitwise_or", False),

This will fail at import time because:

1. BitwiseOrNode is not imported in the import block (lines 32-153)
2. BitwiseOrNode does not exist in the FlatBuffers schema (backends/mlx/serialization/schema.fbs)
3. No corresponding handler or schema definition has been added

This line must be removed from this PR. If bitwise_or support is desired, it needs its own PR with the full schema definition, code generation, import, and tests.

Fix this ->

---

_hardtanh_handler review (backends/mlx/ops.py:2855-2885)

Correctness: The decomposition hardtanh(x, min_val, max_val) = clamp(x, min_val, max_val) is correct, and reusing ClipNode is the right approach.

Issue 1 — Silent fallback on missing metadata (line 2870)

dtype = x_meta.dtype if x_meta is not None else torch.float32

The adjacent _clamp_handler (line 2824-2825) and _relu_handler (line 2684-2685) both raise ValueError when metadata is missing rather than silently defaulting. This handler should follow the same pattern for consistency and to avoid subtle bugs if metadata is unexpectedly absent:

x_meta = n.args[0].meta.get("val")
if x_meta is None:
    raise ValueError("Input tensor metadata not found for hardtanh")
dtype = x_meta.dtype

Fix this ->

Issue 2 — Verbose docstring (lines 2857-2861)

Per the repo's CLAUDE.md, code should be self-documenting with minimal comments. The single-line decorator + function name already conveys what this does. The existing _clamp_handler docstring is somewhat longer, but its semantics are less obvious (optional min/max, None handling). For hardtanh, a one-liner would suffice or the docstring could be removed entirely since the decomposition is trivial.

This is a nit — not blocking.

---

Test coverage (backends/mlx/test/test_ops.py:4103)

The test uses the unary op test harness, which exercises the default min_val=-1, max_val=1 behavior via torch.nn.functional.hardtanh (which uses default arguments). This is reasonable for a first pass, but note that custom min_val/max_val values are not tested. If the handler is intended to support non-default bounds (which it is, given the argument parsing), a dedicated test case with custom bounds would strengthen coverage.

This is minor — not blocking.

---

Summary

Item	Severity
BitwiseOrNode line must be removed (breaks import)	Blocking
Missing metadata should raise, not silently default	Medium
Verbose docstring	Nit
No test for custom min/max bounds	Minor

The core hardtanh implementation is correct and well-structured. The main blocker is the unrelated bitwise_or line that will cause a runtime failure.

[metascroy]

Overall looks great @Jah-yee! Can you address the comment on testing the min/max value bounds?

Also copy/paste the output of your test run (python -m executorch.backends.mlx.test.run_all_tests -k hardtanh) in PR summary.

[metascroy]

Add test cases for min/max value, and paste test output in PR summary :)

[nil-is-all]

@Jah-yee, are you opening another PR to address this?