Llama 3.1 405B
Código abertopor Meta · Lançado em 2024-07-16
38.0
pontuação média
N/A
Preço de entrada
N/A
Preço de saída
N/A
Janela de contexto
text-generation
Tipo
Tested on 21 benchmarks with 38.0% average. Top scores: ARC AI2 (93.7%), HellaSwag (85.6%), TriviaQA (82.7%).
Pontuações de benchmark
| Benchmark | Categoria | Pontuação | Bar |
|---|---|---|---|
| ARC AI2 | knowledge | 93.7 | |
| HellaSwag | knowledge | 85.6 | |
| TriviaQA | knowledge | 82.7 | |
| MMLU | knowledge | 79.3 | |
| Winogrande | knowledge | 78.4 | |
| BBH | reasoning | 77.2 | |
| PIQA | knowledge | 71.8 | |
| MATH level 5 | math | 49.8 | |
| GPQA diamond | knowledge | 34.5 | |
| OpenBookQA | knowledge | 32.3 | |
| MMLU-PRO | knowledge | 25.7 | |
| WeirdML | coding | 21.4 | |
| IFEval | language | 18.1 | |
| OTIS Mock AIME 2024-2025 | math | 9.6 | |
| BBH (HuggingFace) | general | 7.8 | |
| SimpleBench | reasoning | 7.6 | |
| Cybench | coding | 7.5 | |
| The Agent Company | agentic | 7.4 | |
| GPQA | knowledge | 5.9 | |
| MUSR | reasoning | 2.2 | |
| MATH Level 5 | math | 0.0 |
