lambda-qwen2.5-14B-dpo-test: benchmark results
tanliboy's experimental DPO fine-tune of Qwen2.5 14B Instruct on the UltraFeedback binarized dataset. Provider: tanliboy. Released 2024-09-20. Access: Open.
Aggregate Rating 1588 ± 16, rank #481 of 1656 rated models, from 194 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-CKA | 46.25 | Normalized accuracy (%) | 99.8 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 77.71 | Exact match, flexible extract (%) | 99.7 |
| Open LLM Leaderboard - IFEval | 82.31 | Score | 99 |
| Open LLM Leaderboard - MATH Level 5 | 54.61 | Score | 99 |
| Open Korean LLM - KorNAT-Helpful | 66.4 | Normalized accuracy (%) | 98.6 |
| Open Ko-LLM Leaderboard | 52.18 | Average (%) | 98.5 |
| Open Korean LLM - Ko-IFEval | 66.72 | Strict accuracy, prompt/instruction mean (%) | 98.5 |
| Open Korean LLM Leaderboard | 52.18 | Average Score (%) | 98.5 |
| Open Arabic LLM - Aratrust Unfairness | 96.36 | Accuracy (%) | 98.4 |
| Open Arabic LLM - Arabic MMLU Arabic Language (Middle School) | 85.19 | Accuracy (%) | 98.1 |
| Open Korean LLM - KorNAT-Harmless | 77.32 | Normalized accuracy (%) | 98 |
| Open Arabic LLM - Arabic MMLU Computer Science (University) | 81.25 | Accuracy (%) | 97.8 |
Interactive version: theaggregate.ai/model?slug=lambda-qwen2-5-14b-dpo-test · How It Works · Data refreshed daily, snapshot 2026-10-20.