about base model

[manmushanhe]

In your paper, it was stated that you used Qwen2-VL for fine-tuning. On github, it was said that ui-tars was used for fine-tuning. Which model was used for fine-tuning?

[Tongyu-Yan]

In your paper, it was stated that you used Qwen2-VL for fine-tuning. On github, it was said that ui-tars was used for fine-tuning. Which model was used for fine-tuning?

UI-TARS fine-tuned the Qwen2-VL and Qwen2.5-VL models.