Grid Navigation on FrozenLake v1.0 (Drift I)
7,500Success RateVanilla + GLOVE
Evaluation Results
| Method | Links | |
|---|---|---|
| Vanilla + GLOVEBackbone=DeepSeek-V3.22026.01 | 7,500 | |
| MemoryBank + GLOVEBackbone=DeepSeek-V3.22026.01 | 7,000 | |
| Voyager + GLOVEBackbone=DeepSeek-V3.22026.01 | 6,500 | |
| Generative Agent + GLOVEBackbone=DeepSeek-V3.22026.01 | 6,500 | |
| VoyagerBase LLM=Grok-3, Agent Architecture=Voyager, GLOVE Augmentation=false, Drift Visibility=Verifier-Visible2026.01 | 95 | |
| Generative Agent + GLOVEBase LLM=Grok-3, Agent Architecture=Generative Agent, GLOVE Augmentation=true, Drift Visibility=Verifier-Visible2026.01 | 85 | |
| Vanilla + GLOVEBase LLM=Grok-3, Agent Architecture=Vanilla, GLOVE Augmentation=true, Drift Visibility=Verifier-Visible2026.01 | 80 | |
| VanillaBase LLM=Grok-3, Agent Architecture=Vanilla, GLOVE Augmentation=false, Drift Visibility=Verifier-Visible2026.01 | 75 | |
| MemoryBankBase LLM=Grok-3, Agent Architecture=MemoryBank, GLOVE Augmentation=false, Drift Visibility=Verifier-Visible2026.01 | 70 | |
| Voyager + GLOVEBase LLM=Grok-3, Agent Architecture=Voyager, GLOVE Augmentation=true, Drift Visibility=Verifier-Visible2026.01 | 70 | |
| MemoryBank + GLOVEBase LLM=Grok-3, Agent Architecture=MemoryBank, GLOVE Augmentation=true, Drift Visibility=Verifier-Visible2026.01 | 65 | |
| No Memory (Plain)Backbone=DeepSeek-V3.22026.01 | 0 | |
| VanillaBackbone=DeepSeek-V3.22026.01 | 0 | |
| MemoryBankBackbone=DeepSeek-V3.22026.01 | 0 | |
| VoyagerBackbone=DeepSeek-V3.22026.01 | 0 | |
| Generative AgentBackbone=DeepSeek-V3.22026.01 | 0 | |
| No Memory (Plain)Base LLM=Grok-3, Agent Architecture=No Memory, GLOVE Augmentation=false, Drift Visibility=Verifier-Visible2026.01 | 0 | |
| Generative AgentBase LLM=Grok-3, Agent Architecture=Generative Agent, GLOVE Augmentation=false, Drift Visibility=Verifier-Visible2026.01 | 0 |