Embodied Task Planning on RLBench Seen domains
71.89Success RateTMoW
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TMoWBackbone=Llama-3.2-1B2026.01 | 71.89 | 6.3 | |
| SayCanPaySay Backbone=Llama-3.2-3B, Pay Backbone=Llama-3.2-1B2026.01 | 68.08 | 8.69 | |
| LLM+FTBackbone=Llama-3.2-1B, Training=Fine-tuned2026.01 | 65.53 | 11.04 | |
| FLAREBackbone=Llama-3.2-3B2026.01 | 53.05 | 14.77 | |
| LLM-PlannerBackbone=Llama-3.2-3B, Adaptation=Few-shot2026.01 | 35.21 | 14.77 | |
| ZSPBackbone=Llama-3.2-3B, Adaptation=Zero-shot2026.01 | 11.26 | 17.57 |