---
dataset_info:
  features:
  - name: data_source
    dtype: string
  - name: prompt
    list:
    - name: content
      dtype: string
    - name: role
      dtype: string
  - name: images
    list: image
  - name: ability
    dtype: string
  - name: env_name
    dtype: string
  - name: reward_model
    struct:
    - name: ground_truth
      dtype: string
    - name: style
      dtype: string
  - name: extra_info
    struct:
    - name: answer
      dtype: string
    - name: comparison_type
      dtype: string
    - name: count
      dtype: int64
    - name: curriculum_stage
      dtype: int64
    - name: ground_truth_points
      dtype: string
    - name: index
      dtype: string
    - name: original_question
      dtype: string
    - name: point_coordinates
      dtype: string
    - name: question
      dtype: string
    - name: reference_point
      dtype: string
    - name: reference_x
      dtype: float64
    - name: reference_y
      dtype: float64
    - name: split
      dtype: 'null'
    - name: x_unit_step
      dtype: float64
    - name: y_unit_step
      dtype: float64
  splits:
  - name: train
    num_bytes: 86242522
    num_examples: 4000
  download_size: 74868257
  dataset_size: 86242522
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---