Run1 output formatting pareto results
Format and write Pareto frontier results to CSV file with proper rounding and column ordering.From its SKILL.md
npx -y skills add cxcscmu/SkillLearnBench --skill run1_output-formatting-pareto-resultsAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
SKILL.md
1.3 KB, 353 tokens by cl100k_base, as published. Nobody here has run it
Output File
Write results to: /root/pareto_frontier.csv
CSV Format
F1,delta,min_samples,epsilon,shape_weight
Column order (exactly as shown):
F1— F1 score (float)delta— Average Euclidean distance of matches (float)min_samples— DBSCAN parameter (integer)epsilon— DBSCAN parameter (integer)shape_weight— Custom distance weight (float)
Rounding Requirements
F1: Round to 5 decimal placesdelta: Round to 5 decimal placesmin_samples: Integer (no decimal)epsilon: Integer (no decimal)shape_weight: Round to 1 decimal place
Row Ordering
Sort rows by:
- F1 (descending) — highest F1 first
- delta (ascending) — lowest delta second (for ties in F1)
Example Format
F1,delta,min_samples,epsilon,shape_weight
0.75234,12.34567,5,12,1.2
0.74891,13.45678,6,14,1.3
0.72345,11.23456,4,10,1.1
Handling Special Values
- If delta is NaN for a Pareto solution (no matches found), it should not be included in final results (it would be filtered out during F1 > 0.5 check or during Pareto detection)
- All reported solutions should have valid (non-NaN) F1 and delta values
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.