LIMA-13B-hf: benchmark results
heegyu's LIMA 13B, a Transformers/HuggingFace-compatible conversion of Meta's original (v1) LLaMA 13B base model. Provider: Other. Released 2023-08-01. Access: Open.
Aggregate Rating 1300 ± 15, rank #8737 of 10916 rated models, from 77 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - College Mathematics | 38 | Accuracy (%) | 89.5 |
| MMLU-by-task - US Foreign Policy | 82 | Accuracy (%) | 79 |
| MMLU-by-task - Formal Logic | 36.51 | Accuracy (%) | 76.5 |
| MMLU-by-task - Abstract Algebra | 33 | Accuracy (%) | 75.6 |
| MMLU-by-task - HellaSwag | 61.98 | Accuracy (%) | 73.3 |
| MMLU-by-task - Human Sexuality | 61.83 | Accuracy (%) | 71 |
| MMLU-by-task - High School World History | 72.15 | Accuracy (%) | 67.6 |
| MMLU-by-task - ARC Challenge | 54.27 | Accuracy (%) | 66 |
| MMLU-by-task - Astronomy | 51.32 | Accuracy (%) | 65.2 |
| MMLU-by-task - Global Facts | 35 | Accuracy (%) | 64.2 |
| MMLU-by-task - Professional Medicine | 50.74 | Accuracy (%) | 62.7 |
| MMLU-by-task - High School Computer Science | 51 | Accuracy (%) | 62.1 |
Interactive version: theaggregate.ai/model?slug=lima-13b-hf · How It Works · Data refreshed daily, snapshot 2026-11-08.