Embodied Interaction on ALFWorld
94.78Success RateNeuro-Symbolic Dual Memory Framework
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Neuro-Symbolic Dual Memory Framework2026.04 | 94.78 | — | — | |
| AWM2026.04 | 88.81 | — | — | |
| ExpeL2026.04 | 85.07 | — | — | |
| WALL-E 2.02026.04 | 82.84 | — | — | |
| Reflexion2026.04 | 82.66 | — | — | |
| ReAct2026.04 | 76.87 | — | — | |
| ADaPT2026.04 | 72.39 | — | — | |
| StateAct2026.04 | 63.43 | — | — | |
| AgeMemLLM Backbone=Qwen3-4B-Instruct2026.01 | 48.97 | 54.31 | — | |
| Mem0LLM Backbone=Qwen3-4B-Instruct2026.01 | 41.17 | 44.7 | — | |
| AgeMemLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 41.07 | 41.96 | — | |
| LangMemLLM Backbone=Qwen3-4B-Instruct2026.01 | 40.89 | 43.25 | — | |
| No-MemoryLLM Backbone=Qwen3-4B-Instruct2026.01 | 38.51 | 43.97 | — | |
| LangMemLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 38.27 | 34.23 | — | |
| AgeMem-noRLLLM Backbone=Qwen3-4B-Instruct2026.01 | 38.02 | 45.59 | — | |
| AgeMem-noRLLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 37.9 | 33.43 | — | |
| Mem0LLM Backbone=Qwen2.5-7B-Instruct2026.01 | 37.49 | 37.14 | — | |
| Mem0gLLM Backbone=Qwen3-4B-Instruct2026.01 | 36.69 | 41.95 | — | |
| Mem0gLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 35.34 | 36.31 | — | |
| A-MemLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 34.68 | 36.78 | — | |
| A-MemLLM Backbone=Qwen3-4B-Instruct2026.01 | 34.31 | 45.74 | — | |
| No-MemoryLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 27.16 | 28.05 | — | |
| MCTSBase LLM=Llama2-70B2025.07 | — | — | 68.7 | |
| SFTBase LLM=Llama2-70B2025.07 | — | — | 63.4 | |
| Tree DPOBase LLM=Llama2-70B2025.07 | — | — | 65.7 |