Llama 2 13B LoRA Assemble: benchmark results
Provider: Meta. Access: Open.
Unified ELO 1420 ± 10, rank #1974 of 2928 rated models, from 78 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - TruthfulQA MC1 | 40.39 | Accuracy (%) | 96 |
| MMLU-by-task - TruthfulQA MC2 | 55.96 | Accuracy (%) | 94.3 |
| MMLU-by-task - Moral Scenarios | 48.49 | Accuracy (%) | 94.2 |
| MMLU-by-task - College Computer Science | 54 | Accuracy (%) | 93.4 |
| MMLU-by-task - High School Chemistry | 48.77 | Accuracy (%) | 92.1 |
| MMLU-by-task - College Medicine | 59.54 | Accuracy (%) | 91.9 |
| MMLU-by-task - Professional Medicine | 61.03 | Accuracy (%) | 91.6 |
| MMLU-by-task - Electrical Engineering | 56.55 | Accuracy (%) | 91.5 |
| Open Korean LLM - KorNAT-Harmless | 67.92 | Normalized accuracy (%) | 91 |
| MMLU-by-task - High School Government And Politics | 88.08 | Accuracy (%) | 90.9 |
| MMLU-by-task - High School Mathematics | 32.59 | Accuracy (%) | 90.6 |
| MMLU-by-task - High School US History | 83.33 | Accuracy (%) | 90.5 |
Interactive version: theaggregate.ai/model?slug=llama-2-13b-lora-assemble · How It Works · Data refreshed daily, snapshot 2026-09-23.