tulu-2-dpo-13B: benchmark results
Allen AI's 13B chat model, a DPO fine-tune of Llama 2 13B on public, synthetic and human-created data, part of the Tulu V2 suite. Provider: Allen AI. Released 2023-11-12. Access: Open.
Aggregate Rating 1368 ± 22, rank #1340 of 1656 rated models, from 25 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AlpacaEval | 81.24 | Length-Controlled Win Rate (%) | 75.4 |
| RewardBench Chat | 95.81 | Accuracy (%) | 67.6 |
| Open Portuguese LLM - FaQuAD NLI | 72.65 | Macro F1 (%) | 66.5 |
| AlpacaEval 1.0 | 88.12 | Win Rate (%) | 66.3 |
| BiGGen-Bench | 3.42 | Average Score (1-5) | 53.9 |
| Open Portuguese LLM - ASSIN2 RTE | 89.89 | Macro F1 (%) | 52.9 |
| RewardBench Safety | 79.46 | Accuracy (%) | 47.4 |
| Open Portuguese LLM - OAB Exams | 41.09 | Accuracy (%) | 45.8 |
| Open CoT - LSAT Logical Reasoning | 9.41 | CoT Gain (%) | 45 |
| Open Portuguese LLM - ENEM | 58.43 | Accuracy (%) | 44 |
| RewardBench | 73.68 | Score (%) | 43.2 |
| RewardBench Chat Hard | 58.33 | Accuracy (%) | 41.5 |
Interactive version: theaggregate.ai/model?slug=tulu-2-dpo-13b · How It Works · Data refreshed daily, snapshot 2026-10-20.