Pass Rate Evaluation on ALFWorld (Interactive Agent)
74.15Pass@1EARLYEXP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| EARLYEXPBackbone=Llama3.1-8B2026.03 | 74.15 | 95.71 | |
| GRPO-RLVRBackbone=Llama3.1-8B2026.03 | 72.5 | 94.29 | |
| EARLYEXPBackbone=Qwen2.5-7B2026.03 | 72.23 | 92.86 | |
| LEAFEBackbone=Llama3.1-8B2026.03 | 71.79 | 96.43 | |
| ACEBackbone=Llama3.1-8B2026.03 | 70.32 | 92.5 | |
| GRPO-RLVRBackbone=Qwen2.5-7B2026.03 | 69.46 | 91.43 | |
| LEAFEBackbone=Qwen2.5-7B2026.03 | 67.5 | 94.29 | |
| ACEBackbone=Qwen2.5-7B2026.03 | 66.34 | 89.63 | |
| BASE (NO FT)Backbone=Llama3.1-8B2026.03 | 29.82 | 74.29 | |
| BASE (NO FT)Backbone=Qwen2.5-7B2026.03 | 26.07 | 78.57 |