Platypus2-7B: benchmark results
Cole Hunter and Ariel Lee's Platypus2 7B, a LoRA fine-tune of Llama 2 7B on their STEM and logic focused Open-Platypus dataset. Provider: garage-bAInd. Released 2023-08-22. Access: Open.
Aggregate Rating 1306 ± 14, rank #8631 of 10916 rated models, from 77 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - Professional Law | 43.16 | Accuracy (%) | 81.4 |
| MMLU-by-task - High School World History | 75.53 | Accuracy (%) | 79.3 |
| MMLU-by-task - High School European History | 69.09 | Accuracy (%) | 77.8 |
| MMLU-by-task - High School Chemistry | 41.87 | Accuracy (%) | 74.7 |
| MMLU-by-task - Conceptual Physics | 44.68 | Accuracy (%) | 74.5 |
| MMLU-by-task - Econometrics | 31.58 | Accuracy (%) | 73.2 |
| MMLU-by-task - Moral Scenarios | 30.73 | Accuracy (%) | 68.8 |
| MMLU-by-task - International Law | 69.42 | Accuracy (%) | 67 |
| MMLU-by-task - Prehistory | 58.95 | Accuracy (%) | 66.6 |
| MMLU-by-task - Philosophy | 60.45 | Accuracy (%) | 66.4 |
| MMLU-by-task - Professional Psychology | 52.29 | Accuracy (%) | 66.4 |
| MMLU-by-task - Public Relations | 60 | Accuracy (%) | 65.9 |
Interactive version: theaggregate.ai/model?slug=platypus2-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.