Mocha-Dare-7B-ex: benchmark results
Azazelle's DARE-TIES merge, whose published recipe uses mistralai/Mistral-7B-v0.1 as its base and Open-Orca/Mistral-7B-OpenOrca, akjindal53244/Mistral-7B-v0.1-Open-Platypus, WizardLM/WizardMath-7B-V1.1 as merge inputs. Its published Mistral configuration specifies 32 decoder layers. Provider: Azazelle. Access: Open.
Aggregate Rating 1392 ± 33, rank #6466 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - GSM8K | 53.6 | Accuracy (%) (5-shot) | 65.3 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 53.44 | MC2 (%) (0-shot) | 59.5 |
| Open LLM Leaderboard v1 - ARC Challenge | 61.26 | Normalized accuracy (%) (25-shot) | 52 |
| Open LLM Leaderboard v1 - MMLU | 60.77 | Accuracy (%) (5-shot) | 49.6 |
| Open LLM Leaderboard v1 - HellaSwag | 81.6 | Normalized accuracy (%) (10-shot) | 45.9 |
| Open LLM Leaderboard v1 - WinoGrande | 73.8 | Accuracy (%) (5-shot) | 33.6 |
Interactive version: theaggregate.ai/model?slug=mocha-dare-7b-ex · How It Works · Data refreshed daily, snapshot 2026-11-08.