We have identified an issue while reproducing your work.
The model prediction results of 'android_detailed_caption_box' task and 'android_detailed_caption_point' task is been truncated. When I set "max_new_tokens" from 256 to 1024 of 'android_detailed_caption_box' task and 'android_detailed_caption_point' task, the problem has been solved.
Please confirm if need to adjust the parameters of 'android_detailed_caption_box' and 'android_detailed_caption_point' tasks. Look forward to your reply!
We have identified an issue while reproducing your work.
The model prediction results of 'android_detailed_caption_box' task and 'android_detailed_caption_point' task is been truncated. When I set "max_new_tokens" from 256 to 1024 of 'android_detailed_caption_box' task and 'android_detailed_caption_point' task, the problem has been solved.
Please confirm if need to adjust the parameters of 'android_detailed_caption_box' and 'android_detailed_caption_point' tasks. Look forward to your reply!