Procedural Planning on Macro Average In-domain
56.3Macro AccuracyGPT-4o-mini
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-4o-miniModel capacity=Frontier (approx. 700B), Input format=Socratic configuration2026.05 | 56.3 | |
| Gemini-1.5-FlashModel capacity=Frontier (approx. 700B), Input format=Socratic configuration2026.05 | 55.5 | |
| RECIPE-RL 7BModel configuration=7B2026.05 | 46.6 | |
| Qwen2.5-7B (Base)Model configuration=7B2026.05 | 39.2 |