Run2 parallel grid search
Parallel DBSCAN hyperparameter grid search with joblib for Mars cloud clustering.From its SKILL.md
npx -y skills add cxcscmu/SkillLearnBench --skill run2_parallel-grid-searchAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
SKILL.md
0.9 KB, 186 tokens by cl100k_base, as published. Nobody here has run it
Parallel Grid Search
Optimization Tips
- Pre-group data by image outside the parallel loop
- Use
n_jobs=-1for all cores - Each evaluation function should be self-contained
from joblib import Parallel, delayed
from itertools import product
params = list(product(min_samples_range, epsilon_range, shape_weight_range))
results = Parallel(n_jobs=-1)(
delayed(evaluate_params)(ms, eps, sw) for ms, eps, sw in params
)
Greedy Matching for Evaluation
- Compute Euclidean distance matrix between centroids and expert points
- Sort all pairs by distance ascending
- Greedily match closest unmatched pairs (max distance threshold = 100px)
- F1 = 2 * precision * recall / (precision + recall)
- precision = matched / num_centroids
- recall = matched / num_expert_points
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.