paraphrase-multilingual-MiniLM-L12-v2: benchmark results
Provider: sentence-transformers. Released 2021-08-01. Access: Open.
Aggregate Rating 1355 ± 30, rank #7575 of 10916 rated models, from 132 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Romanian NLU - RONEC | 51.66 | Micro F1 excluding MISC (%) | 94.7 |
| SkMTEB - Pair Classification | 67.82 | Mean score (%) over the three pair-classification datasets; | 88.9 |
| EuroEval Bosnian NLU - WikiANN BS | 72.71 | Micro F1 excluding MISC (%) | 82.9 |
| EuroEval Croatian NLU - WikiANN HR | 65.38 | Micro F1 excluding MISC (%) | 78.1 |
| EuroEval Serbian NLU - UNER SR | 75.4 | Micro F1 excluding MISC (%) | 77 |
| EuroEval Hungarian NLU - Szeged NER | 68.13 | Micro F1 excluding MISC (%) | 75.7 |
| EuroEval Latvian NLU - Fullstack NER LV | 55.57 | Micro F1 excluding MISC (%) | 74.6 |
| EuroEval Albanian NLU - WikiANN SQ | 75.2 | Micro F1 excluding MISC (%) | 73.6 |
| EuroEval Croatian NLU - MMS HR | 39.96 | Matthews correlation coefficient (%) | 71.6 |
| SkMTEB - Bitext Mining | 94.91 | Mean F1 (%) over the six bitext-mining datasets; higher is b | 70.4 |
| EuroEval Bulgarian NLU - BG NER Bsnlp | 76.37 | Micro F1 excluding MISC (%) | 70.1 |
| EuroEval Latvian NLU - Latvian Twitter Sentiment | 44.1 | Matthews correlation coefficient (%) | 69 |
Interactive version: theaggregate.ai/model?slug=paraphrase-multilingual-minilm-l12-v2 · How It Works · Data refreshed daily, snapshot 2026-11-08.