Embodied Task Planning on ALFWorld textual observation (seen)
92.5Success RateDynaMind
Evaluation results
| Method | Links | |
|---|---|---|
| DynaMindBase Model=Qwen2.5-7B2026.04 | 92.5 | |
| RoboAgentBase Model=Qwen2.5-VL-3B2026.04 | 92.1 | |
| GiGPOBase Model=Qwen2.5-7B2026.04 | 90.8 | |
| BPOBase Model=Llama3.1-8B2026.04 | 87.9 | |
| SEEA-R1Base Model=Qwen2.5-7B2026.04 | 85.3 | |
| MPOBase Model=Llama3.1-8B2026.04 | 85 | |
| Zero-ShotBase Model=GPT-4o2026.04 | 78.6 | |
| IPRBase Model=Llama-2-7B2026.04 | 70.3 | |
| ETOBase Model=Llama2-7B2026.04 | 68.6 |