Unbabel/M-Prometheus-7B

Unbabel's M-Prometheus-7B is a 7.6 billion parameter language model designed as an open LLM judge for native multilingual output evaluation. Trained on 480k instances of multilingual direct assessment and pairwise comparison data with long-form feedback, it excels at assessing machine translation quality across various languages. This model is specifically optimized for evaluating linguistic nuances and providing detailed feedback on translation accuracy, fluency, and style.