Benchmarks

45 benchmarks in 8 domains

Imaging

4 benchmarks · zero-shot · 4 models

Protocol

Model Organisation Task Min. coverage Hide train/test overlap Sort Order

Benchmarks: 4 of 4

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

CT spleen segmentation (MSD Task09) · Segment spleen ct
Cell Painting knockout retrieval (RxRx3-core) · Retrieve genetic perturbations
Chest X-ray findings (NIH ChestX-ray14) · Classify chest xray findings
Pediatric pneumonia detection (chest X-ray) · Detect pneumonia cxr

Reset

4 models

Ranked from 2 of 4 benchmarks (a third, at least 2)

#	Model ↕	Mean rank ▼	Coverage ↕	CT spleen segmentation (MSD Task09) ↑ Dice (zero-shot) ↕	Cell Painting knockout retrieval (RxRx3-core) ↑ Replicate mAP (zero-shot) ↕	Chest X-ray findings (NIH ChestX-ray14) ↑ Mean ROC-AUC, 5 findings (zero-shot) ↕	Pediatric pneumonia detection (chest X-ray) ↑ ROC-AUC (zero-shot) ↕
1	bio-nlp/biomedclip-pubmedbert-256-vit-base-patch16-224	0.50	3/4	·	0.087 (2)	0.678 (1)	0.909 (1)
Low coverage: below 2 of 4, or nothing to compare with. Listed, not ranked.
—	medical-imaging/vit-chest-xray	1.00	1/4	·	·	·	0.717 (2)
—	cell-imaging/subcellportable	0.00	1/4	·	0.101 (1)	·	·
—	medical-imaging/smaformer-medical-image-segmentation	—	1/4	0.901 (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 zero-shot 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.