Mistral7B-PairRM-SPPO-Iter2: benchmark results
UCLA-AGI's Mistral7B-PairRM-SPPO-Iter2, the second iteration of Self-Play Preference Optimization applied to Mistral 7B Instruct v0.2, generating 5 responses per iteration on synthetic UltraFeedback-derived prompts. Provider: UCLA. Released 2024-05-04. Access: Open.
Aggregate Rating 1381 ± 17, rank #1288 of 1656 rated models, from 29 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Harmless | 67.7 | Normalized accuracy (%) | 90.3 |
| Open Korean LLM - Ko-GPQA-Diamond | 26.77 | Normalized accuracy (%) | 88.9 |
| Open Korean LLM - KorNAT-Helpful | 50.3 | Normalized accuracy (%) | 85.1 |
| Open Chinese LLM - TruthfulQA MC | 60.54 | Accuracy (%) | 83.7 |
| Open Korean LLM - Ko-EQ-Bench | 23.57 | EQ-Bench score | 68.9 |
| Open Korean LLM - Ko-IFEval | 31.19 | Strict accuracy, prompt/instruction mean (%) | 59.1 |
| Open Portuguese LLM - ASSIN2 RTE | 90.56 | Macro F1 (%) | 58.7 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 23.58 | Exact match, flexible extract (%) | 56.5 |
| Open Korean LLM Leaderboard | 35.62 | Average Score (%) | 52.6 |
| Open Ko-LLM Leaderboard | 35.62 | Average (%) | 52.5 |
| Open Portuguese LLM - FaQuAD NLI | 65.5 | Macro F1 (%) | 50.6 |
| Open LLM Leaderboard - MuSR | 40.85 | Score | 49.1 |
Interactive version: theaggregate.ai/model?slug=mistral7b-pairrm-sppo-iter2 · How It Works · Data refreshed daily, snapshot 2026-10-20.