laser-dolphin-mixtral-4x7B-dpo: benchmark results
A mixture-of-experts implementation released by macadeliccc based on cognitivecomputations/dolphin-2.6-mistral-7b-dpo-laser, with Dolphin-style ChatML prompting. The card credits Fernando Fernandes and Eric Hartford for laserRMT. Provider: macadeliccc. Access: Open.
Aggregate Rating 1417 ± 34, rank #5331 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 63.77 | MC2 (%) (0-shot) | 81 |
| Open LLM Leaderboard v1 - HellaSwag | 85.81 | Normalized accuracy (%) (10-shot) | 76.8 |
| Open LLM Leaderboard v1 - ARC Challenge | 64.93 | Normalized accuracy (%) (25-shot) | 68.5 |
| Open LLM Leaderboard v1 - GSM8K | 44.88 | Accuracy (%) (5-shot) | 59 |
| Open LLM Leaderboard v1 - WinoGrande | 77.82 | Accuracy (%) (5-shot) | 58.7 |
| Open LLM Leaderboard v1 - MMLU | 63.04 | Accuracy (%) (5-shot) | 58 |
Interactive version: theaggregate.ai/model?slug=laser-dolphin-mixtral-4x7b-dpo · How It Works · Data refreshed daily, snapshot 2026-11-08.