Agent on MiniWob++ (held-in)
87.12Performance (%)ENVISIONS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ENVISIONSBackbone=LLaMA2-Chat (13B), Approach=Env-guided Self-Training2024.06 | 87.12 | 63.56 | |
| ENVISIONSBackbone=LLaMA2-Chat (7B), Approach=Env-guided Self-Training2024.06 | 85.38 | 58.69 | |
| STaR + Env.Backbone=LLaMA2-Chat (13B), Approach=Env-guided Self-Training2024.06 | 85.15 | 60.6 | |
| Claude-2 + LLaMA2-ChatBackbone=LLaMA2-Chat (13B), Approach=Distill-then-Finetune2024.06 | 84.77 | 57.17 | |
| STaR + Env.Backbone=LLaMA2-Chat (7B), Approach=Env-guided Self-Training2024.06 | 83.71 | 55.91 | |
| Claude-2 + LLaMA2-ChatBackbone=LLaMA2-Chat (7B), Approach=Distill-then-Finetune2024.06 | 82.8 | 53.03 | |
| Iterative SFT+DPOBackbone=LLaMA2-Chat (13B), Approach=Reinforced Self-Training2024.06 | 82.73 | 59.77 | |
| GPT-4 + LLaMA2-ChatBackbone=LLaMA2-Chat (7B), Approach=Distill-then-Finetune2024.06 | 81.14 | 52.46 | |
| GPT-4 + LLaMA2-ChatBackbone=LLaMA2-Chat (13B), Approach=Distill-then-Finetune2024.06 | 80.15 | 56.49 | |
| Iterative SFT+DPOBackbone=LLaMA2-Chat (7B), Approach=Reinforced Self-Training2024.06 | 77.05 | 54.73 | |
| Self-RewardingBackbone=LLaMA2-Chat (13B), Approach=Reinforced Self-Training2024.06 | 74.55 | 52.5 | |
| Self-RewardingBackbone=LLaMA2-Chat (7B), Approach=Reinforced Self-Training2024.06 | 69.39 | 44.22 | |
| LLaMA2-Chat (few-shot)Backbone=LLaMA2-Chat (13B), Mode=few-shot2024.06 | 60 | 38.61 | |
| LLaMA2-Chat (few-shot)Backbone=LLaMA2-Chat (7B), Mode=few-shot2024.06 | 51.14 | 25.69 |