README.md

metadata

dataset_info:
  features:
    - name: action_id
      dtype: string
    - name: obs_before_path
      dtype: string
    - name: obs_after_path
      dtype: string
    - name: obs_before
      dtype: image
    - name: obs_after
      dtype: image
    - name: operation
      dtype: string
    - name: bbox
      struct:
        - name: x
          dtype: float32
        - name: 'y'
          dtype: float32
        - name: width
          dtype: float32
        - name: height
          dtype: float32
    - name: metadata
      struct:
        - name: repr
          dtype: string
        - name: text
          dtype: string
    - name: instruction
      dtype: string
    - name: source
      dtype: string
    - name: platform
      dtype: string
    - name: action_space
      sequence: string
  splits:
    - name: train
      num_bytes: 246119840
      num_examples: 350
  download_size: 221186675
  dataset_size: 246119840
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: mit
task_categories:
  - visual-question-answering
language:
  - en

IDM-Single

IDM-Single is part of the IDMBench in AgentStudio for evaluating agent's ability to learn from Internet-scale videos without action labels. Specifically, this dataset requires VLMs to answer the action transition between two screenshots. Please note that this dataset is a test set rather than a training set. Therefore, please do not use it for training. More details are provided in the project page.