Disable smoothing when smoothing cutoff is 0

[kou]

--smoothing-cutoff 0 didn't disable smoothing. _apply_smoothing() clamps the normalized cutoff to 0.01, so 0 applied a ~1.25 Hz low-pass filter. And load_obs(cutoff=0)/load_action(cutoff=0) fell back to the value set by set_smoothing().

Now 0 disables smoothing and a negative cutoff is rejected by set_smoothing().

[Copilot]

Copilot review overview

🟡 Changes recommended

Per-call negative cutoffs remain accepted, the action path lacks regression coverage, and the description is missing its required issue reference.

Review effort: Balanced
Findings: 2

Open (2)

• Reject negative cutoffs before deciding whether to smooth · New
• Add load_action coverage for zero-cutoff behavior · New

What changed in this PR

Updates smoothing semantics so a zero cutoff disables filtering.

Changes:

• Preserves explicit zero cutoffs for observations and actions.
• Rejects negative configured cutoffs.
• Adds tests and documentation.

PR description: Missing the required first line Fix GH-${ISSUE_NUMBER}.

File	Description
src/​openarm_dataset/​dataset.py	Updates smoothing validation and cutoff handling.
src/​openarm_dataset/​convert.py	Documents the CLI zero-cutoff behavior.
tests/​test_dataset_0_4_0_qpos.py	Tests smoothing disablement and validation.
README.md	Documents zero as disabling smoothing.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Copilot review overview

🟡 Changes recommended

Explicit negative cutoffs passed directly to the loading APIs are silently accepted, and the action-loading path lacks regression coverage.

Review effort: Balanced
Findings: 2 · 1

Open (3)

• Add load_action coverage for zero-cutoff behavior
• Reject negative cutoffs before deciding whether to smooth
• Add required Fix GH-issue reference as the first line · New

[Copilot]

Copilot review overview

🟡 Changes recommended

The documented None contract conflicts with the type annotation, and zero-cutoff action loading lacks regression coverage.

Review effort: Balanced
Findings: 2

Open (2)

• Allow None in set_smoothing cutoff type annotation · New
• Add load_action coverage for zero-cutoff behavior

Resolved since last review (2)

• Reject negative cutoffs before deciding whether to smooth
• Add required Fix GH-issue reference as the first line

[Copilot]

Copilot review overview

🔵 Needs a closer look

The public API annotation conflicts with its documented None behavior, and the PR description lacks the required issue-linking first line.

Review effort: Balanced
Findings: 1

Open (1)

• Allow None in set_smoothing cutoff type annotation

Resolved since last review (1)

• Add load_action coverage for zero-cutoff behavior

[Copilot]

Copilot review overview

🟡 Changes recommended

The pull request metadata does not yet satisfy the repository's submission requirements.

Review effort: Balanced
Findings: None

Resolved since last review (1)

• Allow None in set_smoothing cutoff type annotation

[shokubutsuu]

looks good!