Nous-Hermes-llama-2-7B: benchmark results
Nous Research's Nous-Hermes-Llama2-7b, a Llama 2 7B fine-tune on over 300,000 mostly GPT-4-generated instructions, using the same dataset as the original Llama-1 Hermes for consistency. Provider: Nous Research. Released 2023-07-25. Access: Open.
Aggregate Rating 1301 ± 10, rank #1462 of 1635 rated models, from 91 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - Business Ethics | 57 | Accuracy (%) | 81.2 |
| MMLU-by-task - Professional Medicine | 54.41 | Accuracy (%) | 78.1 |
| MMLU-by-task - Electrical Engineering | 49.66 | Accuracy (%) | 73.9 |
| Hallucinations Leaderboard | 46.65 | Average Task Score (score) | 73.7 |
| MMLU-by-task - TruthfulQA MC1 | 33.41 | Accuracy (%) | 73 |
| MMLU-by-task - Machine Learning | 37.5 | Accuracy (%) | 71.9 |
| MMLU-by-task - TruthfulQA MC2 | 49.01 | Accuracy (%) | 71.6 |
| MMLU-by-task - Abstract Algebra | 32 | Accuracy (%) | 68.1 |
| MMLU-by-task - College Computer Science | 44 | Accuracy (%) | 67.6 |
| Open LLM Leaderboard - MuSR | 42.57 | Score | 66.1 |
| MMLU-by-task - College Medicine | 45.66 | Accuracy (%) | 63.6 |
| MMLU-by-task - Human Aging | 59.64 | Accuracy (%) | 63.6 |
Interactive version: theaggregate.ai/model?slug=nous-hermes-llama-2-7b · How It Works · Data refreshed daily, snapshot 2026-10-15.