Atlas
Sign in

Benchmarks

45 benchmarks in 8 domains

Imaging

3 benchmarks · probe · 6 models
Protocol
Benchmarks: 3 of 3

Mean rank and coverage are recomputed over the benchmarks ticked here. None ticked means all.

Reset

6 models

 Ranked from 2 of 3 benchmarks (a third, at least 2)
# Model ↕ Mean rank ▲ Coverage ↕ Cell Painting knockout retrieval (RxRx3-core) ↑ Knockout accuracy, 20-way (logistic probe) ↕ Chest X-ray findings (NIH ChestX-ray14) ↑ Mean ROC-AUC, 5 findings (logistic probe) ↕ Pediatric pneumonia detection (chest X-ray) ↑ ROC-AUC (logistic probe) ↕
1 medical-imaging/yolov8m-chest-xray-classification overlap 0.25 2/3 · 0.645 (3) 0.992 (1) *
2 bio-nlp/biomedclip-pubmedbert-256-vit-base-patch16-224 0.50 3/3 0.080 (2) 0.689 (1) 0.964 (3)
2 medical-imaging/chest-x-ray-basic overlap 0.50 2/3 · 0.649 (2) * 0.958 (4)
4 medical-imaging/yolov8n-chest-xray-classification overlap 0.63 2/3 · 0.634 (5) 0.989 (2) *
5 medical-imaging/vit-chest-xray 0.88 2/3 · 0.644 (4) 0.954 (5)
Low coverage: below 2 of 3, or nothing to compare with. Listed, not ranked.
— cell-imaging/subcellportable 0.00 1/3 0.120 (1) ··

Each cell is the model's value on that benchmark's board, with its rank there in brackets, as the benchmark's own leaderboard ranks it (its primary probe metric, at its default version); hover for the 95% interval, n and source, click to open that board. A * marks a declared train/test overlap. Rows are ordered by mean normalised rank, (rank − 1) / (board size − 1) averaged over the benchmarks the model has a result on (a board of one entry does not count): ranks, never values, so no average mixes units. Filtering hides rows without re-ranking anyone.