Multi-agent planning and execution on MAT2-THOR Complex Tasks
59TCRScale Plan
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Scale PlanBackbone=GPT-5.22026.03 | 59 | 75 | 92 | |
| LLM+P (PDDL based)Backbone=GPT-5.22026.03 | 24 | 39 | 61 | |
| LaMMA-P (LLM corrected)Backbone=GPT-5.22026.03 | 24 | 56 | 77 | |
| LLM as a PlannerBackbone=GPT-5.22026.03 | 18 | 40 | 69 | |
| LaMMA-P (PDDL plan only)Backbone=GPT-5.22026.03 | 18 | 37 | 66 |