Qwen2 VL 7B Instruct vs Meta Llama 3 8B Instruct
Côte à côte. Chaque métrique. Chaque benchmark.
| Type | Qwen2 VL 7B Instruct | Meta Llama 3 8B Instruct |
|---|---|---|
| Provider |  Alibaba |  Meta |
| score moyen | 47.3 | 45.2 |
| Prix d'entrée | - | - |
| Prix de sortie | - | - |
| Fenêtre de contexte | - | - |
| Sorti le | 2024-08-28 | 2024-04-17 |
| Code source ouvert | Open Source | Open Source |
Scores de benchmark
11 benchmarks · Qwen2 VL 7B Instruct: 10, Meta Llama 3 8B Instruct: 1
| Benchmark | Catégorie | Qwen2 VL 7B Instruct | Meta Llama 3 8B Instruct |
|---|---|---|---|
| BBH (HuggingFace) | general | 35.9 | 28.2 |
| GPQA | knowledge | 9.3 | 1.2 |
| IFEval | language | 46.0 | 74.1 |
| JCommonsenseQA | language | 87.8 | 87.7 |
| JMMLU | language | 56.3 | 46.7 |
| JNLI | language | 74.4 | 61.1 |
| JSQuAD | language | 89.9 | 89.5 |
| LLM-JP — Overall | language | 53.0 | 49.6 |
| MATH Level 5 | math | 19.9 | 8.7 |
| MMLU-PRO | knowledge | 34.4 | 29.6 |
| MUSR | reasoning | 13.6 | 1.6 |