mmBERT-base: benchmark results
Provider: Johns Hopkins University. Released 2025-02-01. Access: Open.
Aggregate Rating 1564 ± 34, rank #1489 of 10916 rated models, from 180 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Serbian NLU - UNER SR | 89.96 | Micro F1 excluding MISC (%) | 98.5 |
| EuroEval Romanian NLU - RONEC | 73.47 | Micro F1 excluding MISC (%) | 98.1 |
| EuroEval Portuguese NLU - HAREM | 68.14 | Micro F1 excluding MISC (%) | 97.5 |
| EuroEval Spanish NLU - CoNLL ES | 81.47 | Micro F1 excluding MISC (%) | 97.3 |
| EuroEval Faroese NLU - FONE | 84.83 | Micro F1 excluding MISC (%) | 97.2 |
| EuroEval Italian NLU - MultiNERD IT | 85.22 | Micro F1 excluding MISC (%) | 97.2 |
| EuroEval Croatian NLU - WikiANN HR | 73.2 | Micro F1 excluding MISC (%) | 97.1 |
| EuroEval Hungarian NLU - Szeged NER | 90.34 | Micro F1 excluding MISC (%) | 97 |
| EuroEval French NLU - Eltec | 73.53 | Micro F1 excluding MISC (%) | 96.3 |
| EuroEval Czech NLU - PONER | 63.94 | Micro F1 excluding MISC (%) | 96.1 |
| EuroEval Bosnian NLU - WikiANN BS | 76.56 | Micro F1 excluding MISC (%) | 95.3 |
| EuroEval German NLU - GermEval | 72.4 | Micro F1 excluding MISC (%) | 94.7 |
Interactive version: theaggregate.ai/model?slug=mmbert-base · How It Works · Data refreshed daily, snapshot 2026-11-08.