Trong Canh Nguyen

UITARS prompt for visual grounding only

1

Currently the prompt need a task description and action history. If I want to use UITARS for visual grounding only, is this possible? What is the prompt you used for...

Wrong coordinates for very big screen images

2

Hello, We observed UITARS 1.5 7B giving wrong coordinates for big screen images (3840x2100). When we resized the images to be smaller it worked correctly. What do you recommend for...

<point> vs |box_start|, |box_end|

1

Hi team, I saw 2 different prompts using different tokens `click(start_box='(x1,y1)')` `click(point='x1 y1')` Which one should I use for UITARS 1.5-7B https://huggingface.co/ByteDance-Seed/UI-TARS-1.5-7B? Do you recommend UITARS 1.5-7B over UI-TARS-7B like...

UITARS-1.5B-7B Grounding error and Screenspot pro score reproduction

4

Hi team, we tried to reproduce the reported ScreenSpot Pro score (48) without success, we used the suggested prompt template and postprocessing, got only around 40. Also we saw some...