metadata
dataset_info:
features:
- name: pre_text
dtype: string
- name: image_path
dtype: image
- name: post_text
dtype: string
- name: answer
dtype: int64
- name: game
dtype: string
- name: problem_type
dtype: string
- name: episode_id
dtype: int64
- name: timestamp
dtype: int64
- name: index
dtype: int64
splits:
- name: train
num_bytes: 1689484388.6
num_examples: 12300
download_size: 1154520567
dataset_size: 1689484388.6
configs:
- config_name: default
data_files:
- split: train
path: data/train-*