Evaluation

Benchmark Leaderboards

Reproducible ablations across tabular, unsupervised, vision, and detection tracks. Ranked by primary metric per task type.

seed 42commit 7d3c95f7/6/2026

tabular

logistic_regression

breast_cancer

98.2%

unsupervised

raw_svm

digits

97.5%

detection

yolov8s

hard_hat_workers

65.8% mAP@50

tabular

titanic+breast_cancer+wine · 1,599 samples

Rank	Model	Dataset	Tool	Score	Train
1	logistic_regression	breast_cancer	scikit-learn	98.2%	2 ms
2	svm_rbf	breast_cancer	scikit-learn	97.4%	11 ms
3	random_forest	breast_cancer	scikit-learn	97.4%	92 ms
4	knn	breast_cancer	scikit-learn	96.5%	0 ms
5	naive_bayes	breast_cancer	scikit-learn	96.5%	0 ms
6	gradient_boosting	breast_cancer	scikit-learn	96.5%	208 ms
7	decision_tree	breast_cancer	scikit-learn	92.1%	3 ms
8	svm_rbf	titanic	scikit-learn	83.2%	16 ms
9	logistic_regression	titanic	scikit-learn	79.7%	12 ms
10	gradient_boosting	titanic	scikit-learn	79.0%	50 ms
11	naive_bayes	titanic	scikit-learn	76.9%	0 ms
12	knn	titanic	scikit-learn	76.2%	2 ms
13	random_forest	titanic	scikit-learn	76.2%	72 ms
14	decision_tree	titanic	scikit-learn	74.8%	1 ms
15	linear_regression	wine_quality	scikit-learn	RMSE 0.711	8 ms

Kaggle Titanic; sklearn Breast Cancer; UCI Wine Quality

Updated 7/6/2026commit 7d3c95fDataset source

iris+digits · 1,257 samples

Rank	Model	Dataset	Tool	Score	Train
1	raw_svm	digits	scikit-learn	97.5%	36 ms
2	raw_knn	digits	scikit-learn	96.4%	1 ms
3	pca_knn	digits	scikit-learn	94.4%	12 ms
4	pca_svm	digits	scikit-learn	94.2%	16 ms
5	dbscan	iris	scikit-learn	silhouette 0.505	1 ms
6	kmeans	iris	scikit-learn	silhouette 0.460	6 ms
7	agglomerative	iris	scikit-learn	silhouette 0.447	36 ms

sklearn Iris and Digits datasets

Updated 7/6/2026commit 7d3c95fDataset source

hard_hat_workers

Rank	Model	Dataset	Tool	Score	Train
1	yolov8s	hard_hat_workers	ultralytics	65.8% mAP@50	0 ms
2	yolov8n	hard_hat_workers	ultralytics	61.2% mAP@50	0 ms

Roboflow Hard Hat Workers (Universe)

Updated 7/6/2026commit 7d3c95fDataset source

kiln-benchmark --track all --seed 42

Vision requires pip install kiln-ml[vision]. YOLO training: Colab notebook #4.