Instruction Following on BabyAI
72.56Success RateAgeMem
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AgeMemLLM Backbone=Qwen3-4B-Instruct2026.01 | 72.56 | 54.31 | |
| AgeMemLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 61.42 | 41.96 | |
| A-MemLLM Backbone=Qwen3-4B-Instruct2026.01 | 61.35 | 45.74 | |
| Mem0LLM Backbone=Qwen2.5-7B-Instruct2026.01 | 60.58 | 37.14 | |
| Mem0LLM Backbone=Qwen3-4B-Instruct2026.01 | 60.05 | 44.7 | |
| A-MemLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 58.82 | 36.78 | |
| Mem0gLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 58.78 | 36.31 | |
| Mem0gLLM Backbone=Qwen3-4B-Instruct2026.01 | 57.59 | 41.95 | |
| AgeMem-noRLLLM Backbone=Qwen3-4B-Instruct2026.01 | 57.48 | 45.59 | |
| No-MemoryLLM Backbone=Qwen3-4B-Instruct2026.01 | 55.83 | 43.97 | |
| LangMemLLM Backbone=Qwen3-4B-Instruct2026.01 | 53.8 | 43.25 | |
| LangMemLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 51.34 | 34.23 | |
| No-MemoryLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 50.8 | 28.05 | |
| AgeMem-noRLLLM Backbone=Qwen2.5-7B-Instruct2026.01 | 46.34 | 33.43 |