Run2 pareto optimization
[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.
npx -y skills add cxcscmu/SkillLearnBench --skill run2_pareto-optimizationAssembled 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
Finding the Pareto frontier for multi-objective optimization using dominance conditions.
SKILL.md
1.6 KB, 339 tokens by cl100k_base, as published. Nobody here has run it
Pareto Optimization (Dominance Condition)
When you need to optimize multiple conflicting objectives (e.g., maximizing F1, minimizing error), you can identify the Pareto frontier by explicitly checking if a point is "dominated" by another point.
Installation
Ensure you have numpy and pandas installed:
pip install numpy pandas
Usage
import numpy as np
import pandas as pd
def get_pareto_frontier(df, obj_max_cols, obj_min_cols):
"""
Finds non-dominated rows in a DataFrame.
"""
pareto_mask = np.ones(len(df), dtype=bool)
for i in range(len(df)):
# Assume self is not dominated unless proven otherwise
row_i = df.iloc[i]
# Check against all other rows
better_or_equal = np.ones(len(df), dtype=bool)
strictly_better = np.zeros(len(df), dtype=bool)
for col in obj_max_cols:
better_or_equal &= (df[col] >= row_i[col])
strictly_better |= (df[col] > row_i[col])
for col in obj_min_cols:
better_or_equal &= (df[col] <= row_i[col])
strictly_better |= (df[col] < row_i[col])
# If any other point is better or equal in all, AND strictly better in at least one
if np.any(better_or_equal & strictly_better):
pareto_mask[i] = False
return df[pareto_mask]
# pareto_df = get_pareto_frontier(results_df, ['F1'], ['delta'])