Logical Reasoning Recommendation: GPT-4
Logical Reasoning → GPT-4
Aperçu du cas d’usage
GPT-4 by Openai is a recommended option for the Logical Reasoning use case.
Modèles recommandés
| Rang | Fournisseur | Fenêtre de contexte | Score |
|---|---|---|---|
| #7 | Openai | 8K | 57/100 |
Analyse de performance
Performance aux benchmarks
| Benchmark | Score |
|---|---|
| MMLU | 85.8 |
| HUMANEVAL | 77.6 |
| GSM8K | 91.7 |
| MATH | 43.9 |
| BBH | 80.9 |
| GPQA | 39.2 |
Forces
- MMLU score 85.8, strong knowledge reasoning.
- GSM8K 91.7, robust math reasoning.
Exigences
- 需要 openai 的 API 密钥
- 输入长度须在 8K 上下文窗口内