llama-3-merged-linear: benchmark results
uygarkurt's llama-3-merged-linear, a training-free mergekit linear merge of three Llama 3 8B models selected from the Open LLM Leaderboard. Provider: uygarkurt. Released 2024-05-01. Access: Open.
Aggregate Rating 1467 ± 14, rank #887 of 1635 rated models, from 39 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Medical LLM - PubMedQA | 78.2 | Accuracy (%) | 96.1 |
| Open Korean LLM - KorNAT-Helpful | 52.65 | Normalized accuracy (%) | 93.6 |
| Open LLM Leaderboard v1 - ARC Challenge | 72.78 | Normalized accuracy (%) (25-shot) | 93.5 |
| Open LLM Leaderboard v1 - HellaSwag | 88.51 | Normalized accuracy (%) (10-shot) | 92.5 |
| Open Korean LLM - Ko-EQ-Bench | 39.85 | EQ-Bench score | 91.2 |
| Open LLM Leaderboard v1 - MMLU | 68.22 | Accuracy (%) (5-shot) | 90.9 |
| Open Korean LLM Leaderboard | 44.27 | Average Score (%) | 90.2 |
| Open Medical LLM - MMLU College Biology | 81.25 | Accuracy (%) | 89.6 |
| Open Korean LLM - Ko-GPQA-Diamond | 26.77 | Normalized accuracy (%) | 88.9 |
| Open LLM Leaderboard v1 - GSM8K | 69.14 | Accuracy (%) (5-shot) | 88.1 |
| Open Korean LLM - Ko-IFEval | 41.62 | Strict accuracy, prompt/instruction mean (%) | 85 |
| Open Medical LLM - MMLU Medical Genetics | 83 | Accuracy (%) | 84.1 |
Interactive version: theaggregate.ai/model?slug=llama-3-merged-linear · How It Works · Data refreshed daily, snapshot 2026-11-05.