Action Planning on Mind2Web high-level setting new
49.08AccuracyWebNaviX +R (our)
Evaluation Results
| Method | Links | |
|---|---|---|
| WebNaviX +R (our)Learning Protocol=In-Context Learning, Model=GPT-4o, Vision=Y, PU=Our-PU2024.09 | 49.08 | |
| WebNaviX (our)Learning Protocol=In-Context Learning, Model=GPT-4o, Vision=Y, PU=Our-PU2024.09 | 47.37 | |
| MindActLearning Protocol=Supervised Fine-Tuning, Model=Flan-T5XL, Vision=N, PU=DeBERTa2024.09 | 46.75 | |
| MindActLearning Protocol=Supervised Fine-Tuning, Model=Flan-T5L, Vision=N, PU=DeBERTa2024.09 | 44.1 | |
| SeeActLearning Protocol=In-Context Learning, Model=GPT-4o, Vision=Y, PU=DeBERTa2024.09 | 43.1 | |
| WebNaviX (our)Learning Protocol=In-Context Learning, Model=GPT-4o, Vision=N, PU=Our-PU2024.09 | 40.68 | |
| MindActLearning Protocol=In-Context Learning, Model=GPT-4o, Vision=N, PU=DeBERTa2024.09 | 36.27 | |
| WebNaviX (our)Learning Protocol=In-Context Learning, Model=Lamma3 70B, Vision=N, PU=Our-PU2024.09 | 35.13 | |
| WebVoyLearning Protocol=In-Context Learning, Model=GPT-4o, Vision=Y, PU=JS2024.09 | 32.86 | |
| WebVoyLearning Protocol=In-Context Learning, Model=GPT-4o, Vision=N, PU=JS2024.09 | 27 |