Interactive LLM Agent Task Success on ALFWorld Unseen
90.3Success RateQ-Evolve
Evaluation results
| Method | Links | |
|---|---|---|
| Q-EvolveBackbone=Llama-3-8B-Instruct2026.06 | 90.3 | |
| SFT + MPOBackbone=Llama-3-8B-Instruct2026.06 | 81.3 | |
| SFTBackbone=Llama-3-8B-Instruct2026.06 | 80.6 | |
| ETO + MPOBackbone=Llama-3-8B-Instruct2026.06 | 79.1 | |
| WKMBackbone=Llama-3-8B-Instruct2026.06 | 78.2 | |
| ETOBackbone=Llama-3-8B-Instruct2026.06 | 76.4 | |
| KnowAgentBackbone=Llama-3-8B-Instruct2026.06 | 74.9 |