agentsclimarketplace

Run2 pareto optimization

Skill cxcscmu/SkillLearnBench/skills/b2-self-feedback-claude-opus-4-6/dbscan-parameter-tuning/run2_pareto-optimization

[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.

Install
npx -y skills add cxcscmu/SkillLearnBench --skill run2_pareto-optimization

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

Identifying Pareto-optimal solutions for multi-objective optimization (maximize F1, minimize delta).

SKILL.md

1.0 KB, 215 tokens by cl100k_base, as published. Nobody here has run it

Pareto Frontier for Multi-Objective Optimization

Definition

Point A dominates Point B iff:

  • A.F1 >= B.F1 AND A.delta <= B.delta
  • With at least one strict inequality

A point is Pareto-optimal if no other point dominates it.

Efficient Implementation

def find_pareto(results):
    """results: list of (f1, delta, ...) tuples. Maximize f1, minimize delta."""
    # Sort by F1 descending for efficiency
    indexed = sorted(enumerate(results), key=lambda x: -x[1][0])
    pareto = []
    min_delta = float('inf')

    for idx, r in indexed:
        if r[1] <= min_delta:
            pareto.append(r)
            min_delta = r[1]

    return pareto

This O(n log n) approach works because after sorting by F1 descending, a point is Pareto-optimal iff its delta is less than or equal to the minimum delta seen so far.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.