Computer Use Task Execution on Impress
40.7Success RateACuRL
Evaluation Results
| Method | Links | |
|---|---|---|
| ACuRLTraining Environment=Impress, Iteration=Iter 32026.02 | 40.7 | |
| ACuRLTraining Environment=Impress, Iteration=Iter 22026.02 | 38.8 | |
| ACuRLTraining Environment=Thunderbird, Iteration=Iter 22026.02 | 35.4 | |
| ACuRLTraining Environment=Thunderbird, Iteration=Iter 32026.02 | 35.1 | |
| ACuRLTraining Environment=Impress, Iteration=Iter 12026.02 | 34.2 | |
| ACuRLTraining Environment=Celestia, Iteration=Iter 32026.02 | 34.2 | |
| ACuRLTraining Environment=Celestia, Iteration=Iter 12026.02 | 33.6 | |
| ACuRLTraining Environment=Calc, Iteration=Iter 32026.02 | 33.2 | |
| ACuRLTraining Environment=Writer, Iteration=Iter 22026.02 | 32.9 | |
| ACuRLTraining Environment=Calc, Iteration=Iter 12026.02 | 32.3 | |
| ACuRLTraining Environment=KAlgebra, Iteration=Iter 12026.02 | 32.3 | |
| ACuRLTraining Environment=Calc, Iteration=Iter 22026.02 | 31.7 | |
| ACuRLTraining Environment=Celestia, Iteration=Iter 22026.02 | 31.7 | |
| UI-TARS-1.5-7BIteration=N/A, Note=Base agent2026.02 | 31.1 | |
| ACuRLTraining Environment=Thunderbird, Iteration=Iter 12026.02 | 31.1 | |
| ACuRLTraining Environment=Writer, Iteration=Iter 32026.02 | 30.8 | |
| ACuRLTraining Environment=KAlgebra, Iteration=Iter 32026.02 | 30.8 | |
| ACuRLTraining Environment=Writer, Iteration=Iter 12026.02 | 30.7 | |
| ACuRLTraining Environment=KAlgebra, Iteration=Iter 22026.02 | 29.8 |