Grid World Navigation on FrozenLake reward reversal Implicit Drift (Source)
45ScoreVoyager + GLOVE
Evaluation Results
| Method | Links | |
|---|---|---|
| Voyager + GLOVEBackbone=Qwen2.5-7B2026.01 | 45 | |
| VanillaBackbone=Qwen2.5-7B2026.01 | 42.5 | |
| Vanilla + GLOVEBackbone=Qwen2.5-7B2026.01 | 42.5 | |
| VoyagerBackbone=Qwen2.5-7B2026.01 | 42.5 | |
| Generative AgentBackbone=Qwen2.5-7B2026.01 | 42.5 | |
| Generative Agent + GLOVEBackbone=Qwen2.5-7B2026.01 | 42.5 | |
| MemoryBank + GLOVEBackbone=Qwen2.5-7B2026.01 | 37.5 | |
| MemoryBankBackbone=Qwen2.5-7B2026.01 | 35 | |
| No Memory (Plain)Backbone=Qwen2.5-7B2026.01 | 0 |