Calme-4x7B-MoE-v0.1: benchmark results
MaziyarPanahi's MoE combining four of his Calme-7B Mistral fine-tunes, aimed at generating notably clear, calm and coherent text. Provider: MaziyarPanahi. Released 2024-03-17. Access: Open.
Aggregate Rating 1449 ± 15, rank #1006 of 1656 rated models, from 35 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-CKA | 36.05 | Normalized accuracy (%) | 96.2 |
| Open LLM Leaderboard v1 - GSM8K | 71.11 | Accuracy (%) (5-shot) | 95.2 |
| Open Korean LLM - KorNAT-Harmless | 71.7 | Normalized accuracy (%) | 94.6 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 72.86 | MC2 (%) (0-shot) | 94 |
| Open LLM Leaderboard v1 - HellaSwag | 88.6 | Normalized accuracy (%) (10-shot) | 92.9 |
| Open LLM Leaderboard v1 - ARC Challenge | 72.53 | Normalized accuracy (%) (25-shot) | 92.4 |
| Open LLM Leaderboard v1 - WinoGrande | 83.5 | Accuracy (%) (5-shot) | 89.2 |
| Open Chinese LLM - TruthfulQA MC | 61.25 | Accuracy (%) | 87.9 |
| Open Korean LLM Leaderboard | 41.1 | Average Score (%) | 80.7 |
| Open Ko-LLM Leaderboard | 41.1 | Average (%) | 80.6 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 44.43 | Exact match, flexible extract (%) | 79.6 |
| Open Portuguese LLM - ASSIN2 RTE | 92.08 | Macro F1 (%) | 74.2 |
Interactive version: theaggregate.ai/model?slug=calme-4x7b-moe-v0-1 · How It Works · Data refreshed daily, snapshot 2026-10-20.