Meta Llama 3 8B vs Qwen2.5 72B Instruct Abliterated
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | Meta Llama 3 8B | Qwen2.5 72B Instruct Abliterated |
|---|---|---|
| Provider |  Meta | HA HuiHui AI |
| puntuación promedio | 44.2 | 48.1 |
| Precio de entrada | - | - |
| Precio de salida | - | - |
| Ventana de contexto | - | - |
| Publicado el | 2024-04-17 | 2024-10-26 |
| Código abierto | Open Source | Proprietary |
Puntuaciones de benchmark
6 benchmarks · Meta Llama 3 8B: 2, Qwen2.5 72B Instruct Abliterated: 4
| Benchmark | Categoría | Meta Llama 3 8B | Qwen2.5 72B Instruct Abliterated |
|---|---|---|---|
| BBH (HuggingFace) | general | 48.7 | 60.5 |
| GPQA | knowledge | 19.7 | 19.4 |
| IFEval | language | 16.0 | 85.9 |
| MATH Level 5 | math | 18.6 | 60.1 |
| MMLU-PRO | knowledge | 41.2 | 50.4 |
| MUSR | reasoning | 16.0 | 12.3 |