Mahou-1.1-llama3-8B: benchmark results
A DPO fine-tune released by flammen.ai of flammenai/Mahou-1.0-llama3-8B, using flammenai/Grill-preprod-v1_chatML. The release targets casual conversation and character roleplay. Its card specifies ChatML prompting. Provider: flammen.ai. Access: Open.
Aggregate Rating 1456 ± 34, rank #3515 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - GSM8K | 70.66 | Accuracy (%) (5-shot) | 93.8 |
| Open LLM Leaderboard v1 - MMLU | 68.46 | Accuracy (%) (5-shot) | 91.5 |
| Open LLM Leaderboard v1 - ARC Challenge | 70.22 | Normalized accuracy (%) (25-shot) | 85.4 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 60.43 | MC2 (%) (0-shot) | 75.9 |
| Open LLM Leaderboard v1 - HellaSwag | 85.64 | Normalized accuracy (%) (10-shot) | 75.7 |
| Open LLM Leaderboard v1 - WinoGrande | 78.3 | Accuracy (%) (5-shot) | 62 |
Interactive version: theaggregate.ai/model?slug=mahou-1-1-llama3-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.