janus-dpo-7B: benchmark results
KAIST's Janus-DPO-7B, a Mistral 7B v0.2 fine-tune DPO-aligned on a 196K-system-message preference dataset for personalized, diverse-preference alignment. Provider: KAIST. Released 2024-04-25. Access: Open.
Aggregate Rating 1383 ± 30, rank #1281 of 1656 rated models, from 19 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - MuSR | 43.87 | Score | 77.3 |
| Open Portuguese LLM - FaQuAD NLI | 71.47 | Macro F1 (%) | 63.5 |
| Open Portuguese LLM - OAB Exams | 43.19 | Accuracy (%) | 54.4 |
| Open Portuguese LLM - BLUEX | 52.02 | Accuracy (%) | 51.4 |
| Open Portuguese LLM - ENEM | 61.65 | Accuracy (%) | 50.2 |
| Open Portuguese LLM - ASSIN2 RTE | 88.35 | Macro F1 (%) | 44.5 |
| Open Chinese LLM - TruthfulQA MC | 52.92 | Accuracy (%) | 43.8 |
| Open LLM Leaderboard - BBH | 47.73 | Score | 40.8 |
| Open LLM Leaderboard - IFEval | 40.03 | Score | 39.8 |
| Open LLM Leaderboard - GPQA | 28.19 | Score | 36.8 |
| Open LLM Leaderboard - MMLU-Pro | 29.76 | Score | 35.6 |
| Open Chinese LLM - GSM8K | 31.77 | Accuracy (%) | 26.3 |
Interactive version: theaggregate.ai/model?slug=janus-dpo-7b · How It Works · Data refreshed daily, snapshot 2026-10-20.