tulu-2-13B: benchmark results
Allen AI's Tulu 2 13B, a supervised fine-tune of Llama 2 13B on a mix of public, synthetic and human-created datasets, part of the Tulu V2 suite. Provider: Allen AI. Released 2023-11-13. Access: Open.
Aggregate Rating 1341 ± 34, rank #1404 of 1656 rated models, from 22 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LLM Trustworthy - Fairness | 97.9 | Trust Score (%) | 75 |
| LLM Trustworthy - Adversarial Demo | 71.17 | Trust Score (%) | 70.8 |
| LLM Trustworthy - Out-of-Distribution | 70.17 | Trust Score (%) | 62.5 |
| LLM Trustworthy - Stereotype | 89.33 | Trust Score (%) | 54.2 |
| AI-Secure LLM Trustworthy Leaderboard | 0.67 | Trustworthy average | 50 |
| LLM Trustworthy - Toxicity | 44.8 | Trust Score (%) | 50 |
| LLM Trustworthy Leaderboard | 66.51 | Average Trust Score (%) | 50 |
| Open Portuguese LLM - FaQuAD NLI | 61.82 | Macro F1 (%) | 46.2 |
| Open CoT - LogiQA 2 | 5.47 | CoT Gain (%) | 42 |
| LLM Trustworthy - Privacy | 78.9 | Trust Score (%) | 41.7 |
| Open Portuguese LLM - OAB Exams | 39.45 | Accuracy (%) | 41.3 |
| BiGGen-Bench | 3.21 | Average Score (1-5) | 41.2 |
Interactive version: theaggregate.ai/model?slug=tulu-2-13b · How It Works · Data refreshed daily, snapshot 2026-10-20.