Embodied Task Planning on ALFWorld textual observation (unseen)
94Success RateRoboAgent
Evaluation results
| Method | Links | |
|---|---|---|
| RoboAgentBase Model=Qwen2.5-VL-3B2026.04 | 94 | |
| BPOBase Model=Llama3.1-8B2026.04 | 89.6 | |
| DynaMindBase Model=Qwen2.5-7B2026.04 | 89.1 | |
| SEEA-R1Base Model=Qwen2.5-7B2026.04 | 85.1 | |
| LACBase Model=Gemma-7B2026.04 | 84 | |
| Zero-ShotBase Model=GPT-4o2026.04 | 83.6 | |
| MPOBase Model=Llama3.1-8B2026.04 | 79.1 | |
| IPRBase Model=Llama-2-7B2026.04 | 74.7 | |
| ETOBase Model=Llama2-7B2026.04 | 72.4 |