free-evo-qwen72B-v0.8-re: benchmark results
Freewheelin's evolutionary merge of Qwen2-based 72B models, inspired by Sakana AI's evolutionary model-merging method. Provider: Freewheelin. Released 2024-05-02. Access: Open.
Aggregate Rating 1602 ± 19, rank #435 of 1656 rated models, from 143 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - ARC Challenge | 79.86 | Normalized accuracy (%) (25-shot) | 100 |
| Open Chinese LLM - ARC Challenge | 71.16 | Accuracy (%) | 99.9 |
| Open LLM Leaderboard v1 - HellaSwag | 91.34 | Normalized accuracy (%) (10-shot) | 99.9 |
| Open Chinese LLM Leaderboard | 74.62 | Official Nonzero-Task Mean (%) | 99.7 |
| Open LLM Leaderboard v1 - WinoGrande | 87.77 | Accuracy (%) (5-shot) | 99.7 |
| Open Chinese LLM - HellaSwag | 78.09 | Accuracy (%) | 99.4 |
| Open LLM Leaderboard v1 - GSM8K | 75.89 | Accuracy (%) (5-shot) | 99.3 |
| Open LLM Leaderboard v1 - MMLU | 78 | Accuracy (%) (5-shot) | 99.1 |
| Open Arabic LLM - Arabic MMLU HT Global Facts | 54 | Accuracy (%) | 98.8 |
| Open Arabic LLM - Madinah QA Arabic Language (Grammar) | 70.96 | Accuracy (%) | 98.8 |
| Open Chinese LLM - CMMLU | 73.88 | Accuracy (%) | 98.8 |
| Open Chinese LLM - C-Eval Semantic | 91.21 | Accuracy (%) | 98.5 |
Interactive version: theaggregate.ai/model?slug=free-evo-qwen72b-v0-8-re · How It Works · Data refreshed daily, snapshot 2026-10-20.