Llama 2 13B Base: benchmark results
Meta's Llama 2 13B, the base pretrained checkpoint released under the Llama 2 Community License in July 2023; the linked model card provides no further details here. Provider: Meta. Released 2023-07-18. Access: Open.
Aggregate Rating 1319 ± 13, rank #1445 of 1656 rated models, from 155 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MERA Original Paper - ruModAr | 48.6 | Exact Match (%; greedy generation) | 94.4 |
| EuroEval Portuguese NLU - MultiWikiQA PT | 76.05 | Character F1 (%) | 88.9 |
| MERA Original Paper | 36.8 | Total Score (%; mean of 17 problem-solving and exam tasks, M | 88.9 |
| MERA Original Paper - ruMMLU | 56.3 | Accuracy (%; 5-shot, log-likelihood) | 88.9 |
| MERA Original Paper - ruOpenBookQA | 63.7 | Accuracy (%; 5-shot, log-likelihood) | 88.9 |
| MERA Original Paper - ruWorldTree | 70.3 | Accuracy (%; 5-shot, log-likelihood) | 88.9 |
| Hallucinations Leaderboard | 49.78 | Average Task Score (score) | 86.8 |
| MERA Original Paper - MathLogicQA | 31.4 | Accuracy (%; 5-shot, log-likelihood) | 83.3 |
| MERA Original Paper - SimpleAr | 91.1 | Exact Match (%; 5-shot, greedy generation) | 83.3 |
| MERA Original Paper - ruMultiAr | 15.6 | Exact Match (%; 5-shot, greedy generation) | 83.3 |
| EuroEval Italian NLU - SQuAD IT | 71.05 | Character F1 (%) | 76.8 |
| EuroEval Dutch NLU - DBRD | 90.19 | Matthews correlation coefficient (%) | 72.6 |
Interactive version: theaggregate.ai/model?slug=llama-2-13b-base · How It Works · Data refreshed daily, snapshot 2026-10-20.