metadata
dataset_info:
features:
- name: episode_id
dtype: string
- name: actions
dtype: int64
- name: text_actions
dtype: string
- name: plans
dtype: string
- name: rewards
dtype: float64
- name: terms
dtype: bool
- name: truncs
dtype: bool
- name: image_paths
dtype: image
- name: short_term_context
dtype: string
- name: long_term_context
dtype: string
- name: episode_index
dtype: int64
splits:
- name: train
num_bytes: 2138523791
num_examples: 686821
download_size: 1716380280
dataset_size: 2138523791
configs:
- config_name: default
data_files:
- split: train
path: data/train-*