1 benchmark · 54 results. Each benchmark shows the top of its default board (its first primary metric at its default version); ranks are the server's, ties share a rank, and "best" is the best visible result, not a state of the art.
BC5CDR chemical NER bio-nlp/bc5cdr-chemical-ner
Chemical entity F1 (token probe) · ↑ higher is better · probe · atlas-n100-v1
| # | Model | Value [95% CI] | Source | Overlap |
|---|---|---|---|---|
| 1 | bio-nlp/biomedvlp-cxr-bert-specialized | 0.889 [0.857, 0.921] best | reproduced | overlapTrain/test overlap declared by the evaluator: PubMed abstracts |
| 2 | bio-nlp/biobert-base-cased-v1-2 | 0.878 [0.845, 0.912] | reproduced | overlapTrain/test overlap declared by the evaluator: PubMed abstracts |
| 3 | bio-nlp/biomednlp-biomedbert-base-uncased-abstract-fulltext | 0.878 [0.841, 0.912] | reproduced | overlapTrain/test overlap declared by the evaluator: PubMed abstracts |
| 4 | bio-nlp/scibert-scivocab-uncased | 0.874 [0.833, 0.908] | reproduced | unknown |
| 5 | bio-nlp/clinical-assertion-negation-bert | 0.863 [0.823, 0.897] | reproduced | overlapTrain/test overlap declared by the evaluator: PubMed abstracts |
5 of 16 on this board. Full leaderboard