DARE_TIES_13B: benchmark results
yunconglong's DARE TIES merge of two 13B DPO-tuned FusionNet variants, built with mergekit on a Truthful DPO FusionNet MoE base. Provider: yunconglong. Released 2024-01-30. Access: Open.
Aggregate Rating 1478 ± 24, rank #851 of 1635 rated models, from 19 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - WinoGrande | 88.08 | Accuracy (%) (5-shot) | 99.8 |
| Open LLM Leaderboard v1 - HellaSwag | 89.5 | Normalized accuracy (%) (10-shot) | 99.4 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 78.66 | MC2 (%) (0-shot) | 99.3 |
| Open LLM Leaderboard v1 - ARC Challenge | 74.32 | Normalized accuracy (%) (25-shot) | 99.2 |
| Open Chinese LLM - TruthfulQA MC | 61.84 | Accuracy (%) | 89.3 |
| Open Chinese LLM - WinoGrande | 67.64 | Accuracy (%) | 84.6 |
| Open LLM Leaderboard v1 - GSM8K | 67.55 | Accuracy (%) (5-shot) | 83.8 |
| Open Portuguese LLM - FaQuAD NLI | 77.84 | Macro F1 (%) | 81.5 |
| Open Portuguese LLM - ASSIN2 RTE | 92.52 | Macro F1 (%) | 80 |
| Open LLM Leaderboard v1 - MMLU | 64.47 | Accuracy (%) (5-shot) | 70.9 |
| Open Chinese LLM - ARC Challenge | 54.52 | Accuracy (%) | 65.8 |
| Open Chinese LLM - HellaSwag | 62 | Accuracy (%) | 65.4 |
Interactive version: theaggregate.ai/model?slug=dare-ties-13b · How It Works · Data refreshed daily, snapshot 2026-11-05.