Qwen2 VL 7B Instruct vs Qwen2 7B Instruct
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | Qwen2 VL 7B Instruct | Qwen2 7B Instruct |
|---|---|---|
| Provider |  Alibaba | Alibaba |
| puntuación promedio | 47.3 | 50.5 |
| Precio de entrada | - | - |
| Precio de salida | - | - |
| Ventana de contexto | - | - |
| Publicado el | 2024-08-28 | 2024-06-04 |
| Código abierto | Open Source | Open Source |
Puntuaciones de benchmark
11 benchmarks · Qwen2 VL 7B Instruct: 5, Qwen2 7B Instruct: 6
| Benchmark | Categoría | Qwen2 VL 7B Instruct | Qwen2 7B Instruct |
|---|---|---|---|
| BBH (HuggingFace) | general | 35.9 | 37.8 |
| GPQA | knowledge | 9.3 | 6.4 |
| IFEval | language | 46.0 | 56.8 |
| JCommonsenseQA | language | 87.8 | 89.1 |
| JMMLU | language | 56.3 | 56.5 |
| JNLI | language | 74.4 | 81.3 |
| JSQuAD | language | 89.9 | 89.6 |
| LLM-JP — Overall | language | 53.0 | 51.7 |
| MATH Level 5 | math | 19.9 | 27.6 |
| MMLU-PRO | knowledge | 34.4 | 31.6 |
| MUSR | reasoning | 13.6 | 7.4 |