mistral-nemo-wissenschaft-12B: benchmark results
nbeerbower's ORPO fine-tune of Mistral Nemo Instruct 12B on ScienceQA, aimed at correct science question answering. Provider: nbeerbower. Released 2024-08-12. Access: Open.
Aggregate Rating 1530 ± 21, rank #683 of 1654 rated models, from 20 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM Leaderboard | 50.01 | Average Score (%) | 97.3 |
| Open Korean LLM - Ko-EQ-Bench | 57.38 | EQ-Bench score | 97.1 |
| Open Korean LLM - Ko-Winogrande | 70.24 | Accuracy (%) | 96.8 |
| Open Korean LLM - Ko-IFEval | 55.05 | Strict accuracy, prompt/instruction mean (%) | 96.5 |
| Open Portuguese LLM - FaQuAD NLI | 82.54 | Macro F1 (%) | 95 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 58.23 | Exact match, flexible extract (%) | 92.3 |
| Open Korean LLM - KorNAT-Helpful | 50.35 | Normalized accuracy (%) | 85.4 |
| Open Portuguese LLM - OAB Exams | 53.9 | Accuracy (%) | 84.5 |
| Open Portuguese LLM - ENEM | 72.64 | Accuracy (%) | 83.8 |
| Open Portuguese LLM - ASSIN2 RTE | 92.69 | Macro F1 (%) | 82.2 |
| Open LLM Leaderboard - IFEval | 65.2 | Score | 77.5 |
| Open Portuguese LLM - BLUEX | 60.08 | Accuracy (%) | 77.4 |
Interactive version: theaggregate.ai/model?slug=mistral-nemo-wissenschaft-12b · How It Works · Data refreshed daily, snapshot 2026-10-24.