metadata
dataset_info:
features:
- name: question
dtype: string
- name: answer
dtype: string
- name: choices
list: string
- name: topdown_map
dtype: image
- name: scene_name
dtype: string
- name: trajectory_id
dtype: string
- name: question_type
dtype: string
- name: difficulty
dtype: string
- name: total_frames
dtype: int64
- name: movement_type
dtype: string
- name: metadata
struct:
- name: category
dtype: string
- name: count
dtype: int64
- name: cropped_to_trajectory
dtype: bool
- name: frame_count
dtype: int64
- name: generation_method
dtype: string
- name: has_topdown_map
dtype: bool
- name: platform
dtype: string
- name: task_type
dtype: string
- name: frame_0
dtype: image
- name: frame_1
dtype: image
- name: frame_2
dtype: image
- name: frame_3
dtype: image
- name: frame_4
dtype: image
splits:
- name: train
num_bytes: 77651553424.355
num_examples: 19957
download_size: 78156392693
dataset_size: 77651553424.355
configs:
- config_name: default
data_files:
- split: train
path: data/train-*