tulu-2-dpo-7B: benchmark results
Allen AI's 7B chat model, a DPO fine-tune of Llama 2 7B on public, synthetic and human-created data, part of the Tulu V2 suite. Provider: Allen AI. Released 2023-11-12. Access: Open.
Aggregate Rating 1342 ± 20, rank #1400 of 1656 rated models, from 32 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| RewardBench Chat | 97.49 | Accuracy (%) | 90.1 |
| AlpacaEval | 77.85 | Length-Controlled Win Rate (%) | 71.9 |
| JustEval - Depth | 4.36 | Score (1-5) | 60 |
| Open Portuguese LLM - FaQuAD NLI | 67.68 | Macro F1 (%) | 55 |
| AlpacaEval 1.0 | 84.22 | Win Rate (%) | 54.5 |
| JustEval - Engagement | 4.69 | Score (1-5) | 46.7 |
| JustEval - Helpfulness | 4.64 | Score (1-5) | 46.7 |
| Open Portuguese LLM - ASSIN2 RTE | 88.27 | Macro F1 (%) | 44.2 |
| BiGGen-Bench | 3.28 | Average Score (1-5) | 44.1 |
| RewardBench Safety | 75.27 | Accuracy (%) | 39.8 |
| Open Portuguese LLM - OAB Exams | 38.27 | Accuracy (%) | 37.9 |
| RewardBench | 72.12 | Score (%) | 37.5 |
Interactive version: theaggregate.ai/model?slug=tulu-2-dpo-7b · How It Works · Data refreshed daily, snapshot 2026-10-20.