Object Placement Reasoning on MUSR object (test)
68.9AccuracyInduced
Evaluation results
| Method | Links | |
|---|---|---|
| InducedLLM Backbone=DeepSeek-V32026.06 | 68.9 | |
| Program-of-ThoughtsLLM Backbone=DeepSeek-V32026.06 | 60.4 | |
| Hand-designedLLM Backbone=DeepSeek-V32026.06 | 58.5 | |
| Agent Workflow MemoryLLM Backbone=DeepSeek-V32026.06 | 53.8 | |
| ReActLLM Backbone=DeepSeek-V32026.06 | 50 | |
| Chain-of-ThoughtLLM Backbone=DeepSeek-V32026.06 | 39.6 |