executing-plans
executing-plans
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available
スキルを探す
生産性・共同作業 · SKILL.md
agreement_line_fit
agreement_line_fitFit the probit-space ID-to-OOD agreement line and report agreement-on-the-line diagnostics without OOD labels.
Independent GitHub snapshot · Automated classification · Original documentation governs
スキルの説明
GitHubname: agreement_line_fit description: Fit the probit-space ID-to-OOD agreement line and report agreement-on-the-line diagnostics without OOD labels.
Agreement Line Fit
Use this skill when pairwise agreement statistics are available and you need to check whether agreement-on-the-line plausibly holds.
Inputs
- JSON statistics from
agreement_statistics. - Optional R² threshold.
Outputs
- Slope, intercept, R², residual summaries, and
on_linedecision.
Workflow
- Align pairwise ID/OOD agreement probit values.
- Fit ordinary least squares in probit space.
- Compute R² and residual summaries.
- Treat low R² as a reliability warning for downstream OOD accuracy estimates.
Validation
Run python tests/test_agreement_line_fit.py from this skill directory.
Limitations
The line diagnostic is unlabeled; it cannot prove true OOD accuracy without a separate evaluation harness.