17 examples from 1 videos - UI automation tasks from screen recordings.
video_id: Sequential ID for each video (video_001, video_002, etc.)
step: Step number within that video (0, 1, 2, ...)
system: System prompt for the GUI agent
user: Task instruction + previous actions
assistant: Model's reasoning and action
image: Screenshot of the UI state
from datasets import… See the full description on the dataset page:
https://huggingface.co/datasets/KMH158-QLU/test_dataset_132.
