Llama 3 8B Instruct
Open Sourcevon Meta · Veroeffentlicht 2024-04-18
30.8
Durchschn. Score
$0.03/1M
Eingabepreis
$0.04/1M
Ausgabepreis
8K tokens (~4 books)
Kontextfenster
text
Typ
Tested on 16 benchmarks with 30.8% average. Top scores: Chatbot Arena Elo — Overall (1222.2%), ARC AI2 (77.1%), OpenBookQA (76.8%).
Benchmark-Ergebnisse
| Benchmark | Kategorie | Score | Bar |
|---|---|---|---|
| Chatbot Arena Elo — Overall | arena | 1222.2 | |
| ARC AI2 | knowledge | 77.1 | |
| OpenBookQA | knowledge | 76.8 | |
| TriviaQA | knowledge | 67.7 | |
| MMLU | knowledge | 58.4 | |
| Winogrande | knowledge | 51.4 | |
| ANLI | knowledge | 36.0 | |
| IFEval | language | 24.0 | |
| MUSR | reasoning | 19.9 | |
| BBH (HuggingFace) | general | 18.4 | |
| MMLU-PRO | knowledge | 17.8 | |
| MATH level 5 | math | 6.1 | |
| MATH Level 5 | math | 3.9 | |
| GPQA | knowledge | 2.1 | |
| GPQA diamond | knowledge | 1.4 | |
| OTIS Mock AIME 2024-2025 | math | 0.7 |