metadata
dataset_info:
features:
- name: group_id
dtype: string
- name: sample_id
dtype: string
- name: type
dtype:
class_label:
names:
'0': factual_only
'1': factual_counterfactual
- name: image
dtype: image
- name: ground_truth
dtype: image
- name: bbox
sequence: float32
- name: points
sequence:
sequence: int32
- name: masks
dtype: image
- name: results_vis
dtype: image
splits:
- name: train
num_bytes: 135813391736.76
num_examples: 143740
download_size: 27633397982
dataset_size: 135813391736.76
configs:
- config_name: default
data_files:
- split: train
path: data/train-*