mmBERT-small: benchmark results
Provider: Johns Hopkins University. Released 2025-02-01. Access: Open.
Aggregate Rating 1498 ± 33, rank #2394 of 10916 rated models, from 180 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Romanian NLU - RONEC | 72.15 | Micro F1 excluding MISC (%) | 97.1 |
| EuroEval Portuguese NLU - HAREM | 65.42 | Micro F1 excluding MISC (%) | 96.8 |
| EuroEval Hungarian NLU - Szeged NER | 89.7 | Micro F1 excluding MISC (%) | 96.5 |
| EuroEval Spanish NLU - CoNLL ES | 80.93 | Micro F1 excluding MISC (%) | 96.3 |
| EuroEval Croatian NLU - WikiANN HR | 70.66 | Micro F1 excluding MISC (%) | 96.2 |
| EuroEval Faroese NLU - FONE | 84.25 | Micro F1 excluding MISC (%) | 96 |
| EuroEval Italian NLU - MultiNERD IT | 83.75 | Micro F1 excluding MISC (%) | 95.6 |
| EuroEval Serbian NLU - UNER SR | 86.97 | Micro F1 excluding MISC (%) | 94.6 |
| EuroEval French NLU - Eltec | 72.54 | Micro F1 excluding MISC (%) | 94.4 |
| EuroEval Ukrainian NLU - NER UK | 73.3 | Micro F1 excluding MISC (%) | 94 |
| EuroEval English NLU - CoNLL EN | 84.32 | Micro F1 excluding MISC (%) | 91.9 |
| EuroEval Czech NLU - PONER | 60.48 | Micro F1 excluding MISC (%) | 91.2 |
Interactive version: theaggregate.ai/model?slug=mmbert-small · How It Works · Data refreshed daily, snapshot 2026-11-08.