SlimMelodicMaid: benchmark results
Azazelle's SLERP merge, whose published recipe uses mistralai/Mistral-7B-v0.1 as its base and Azazelle/Silicon-Medley, Azazelle/xDAN-SlimOrca as merge inputs. Its published Mistral configuration specifies 32 decoder layers. The card heading retains another label; the exact repository and its recipe define this publication, without a global name alias. Provider: Azazelle. Access: Open.
Aggregate Rating 1433 ± 34, rank #4540 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - HellaSwag | 86.01 | Normalized accuracy (%) (10-shot) | 78.1 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 60.88 | MC2 (%) (0-shot) | 76.6 |
| Open LLM Leaderboard v1 - ARC Challenge | 67.15 | Normalized accuracy (%) (25-shot) | 75.9 |
| Open LLM Leaderboard v1 - MMLU | 64.75 | Accuracy (%) (5-shot) | 75.9 |
| Open LLM Leaderboard v1 - GSM8K | 60.8 | Accuracy (%) (5-shot) | 73 |
| Open LLM Leaderboard v1 - WinoGrande | 78.61 | Accuracy (%) (5-shot) | 64.3 |
Interactive version: theaggregate.ai/model?slug=slimmelodicmaid · How It Works · Data refreshed daily, snapshot 2026-11-08.