flammen23X-mistral-7B: benchmark results
A DPO fine-tune released by flammen.ai of flammenai/flammen23-mistral-7B, using flammenai/character-roleplay-DPO. The card heading omits the X suffix, while its parent and dataset identify this separate fine-tuning stage; no alias is inferred. Provider: flammen.ai. Access: Open.
Aggregate Rating 1445 ± 35, rank #4006 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 68.6 | MC2 (%) (0-shot) | 87.8 |
| Open LLM Leaderboard v1 - HellaSwag | 86.6 | Normalized accuracy (%) (10-shot) | 82 |
| Open LLM Leaderboard v1 - MMLU | 65.05 | Accuracy (%) (5-shot) | 80.9 |
| Open LLM Leaderboard v1 - ARC Challenge | 68.43 | Normalized accuracy (%) (25-shot) | 80.4 |
| Open LLM Leaderboard v1 - GSM8K | 65.66 | Accuracy (%) (5-shot) | 80.4 |
| Open LLM Leaderboard v1 - WinoGrande | 81.06 | Accuracy (%) (5-shot) | 77 |
Interactive version: theaggregate.ai/model?slug=flammen23x-mistral-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.