TurdusTrixBeagle-DARETIES-7B: benchmark results
jsfs11 release whose published recipe uses DARE-TIES merging with mistralai/Mistral-7B-v0.1 as the base and udkai/Turdus, CultriX/MergeTrix-7B-v2, mlabonne/NeuralBeagle14-7B as additional inputs. Provider: jsfs11. Access: Open.
Aggregate Rating 1469 ± 43, rank #2741 of 10870 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - WinoGrande | 85.16 | Accuracy (%) (5-shot) | 97.5 |
| Open LLM Leaderboard v1 - ARC Challenge | 73.46 | Normalized accuracy (%) (25-shot) | 97.4 |
| Open LLM Leaderboard v1 - HellaSwag | 88.61 | Normalized accuracy (%) (10-shot) | 93 |
| Open LLM Leaderboard v1 - GSM8K | 70.28 | Accuracy (%) (5-shot) | 92.4 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 68.81 | MC2 (%) (0-shot) | 88.1 |
| Open LLM Leaderboard v1 - MMLU | 64.89 | Accuracy (%) (5-shot) | 78 |
Interactive version: theaggregate.ai/model?slug=turdustrixbeagle-dareties-7b · How It Works · Data refreshed daily, snapshot 2026-11-12.