LIMA2-7B-hf: benchmark results
heegyu's LIMA2 7B, Llama 2 7B fine-tuned for 10 epochs on the 64bits/lima dataset in Vicuna format. Provider: heegyu. Released 2023-08-04. Access: Open.
Aggregate Rating 1261 ± 15, rank #9153 of 10916 rated models, from 77 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - College Chemistry | 43 | Accuracy (%) | 83.5 |
| MMLU-by-task - High School Physics | 35.1 | Accuracy (%) | 82 |
| MMLU-by-task - HellaSwag | 62.35 | Accuracy (%) | 75.9 |
| MMLU-by-task - High School Mathematics | 29.26 | Accuracy (%) | 70.6 |
| MMLU-by-task - Formal Logic | 34.13 | Accuracy (%) | 66.2 |
| MMLU-by-task - College Mathematics | 33 | Accuracy (%) | 61.7 |
| MMLU-by-task - Sociology | 66.67 | Accuracy (%) | 60.5 |
| MMLU-by-task - College Medicine | 43.35 | Accuracy (%) | 58 |
| MMLU-by-task - Human Sexuality | 57.25 | Accuracy (%) | 58 |
| MMLU-by-task - Professional Medicine | 49.26 | Accuracy (%) | 57.9 |
| MMLU-by-task - TruthfulQA MC1 | 29.74 | Accuracy (%) | 56.6 |
| MMLU-by-task - College Physics | 25.49 | Accuracy (%) | 53.2 |
Interactive version: theaggregate.ai/model?slug=lima2-7b-hf · How It Works · Data refreshed daily, snapshot 2026-11-08.