metadata
dataset_info:
features:
- name: image
dtype: image
- name: prompt
sequence: string
- name: T5_last_hidden_states
sequence:
sequence:
sequence: float32
- name: objects
sequence: string
splits:
- name: train
num_bytes: 1358102542.875
num_examples: 1001
download_size: 1191026712
dataset_size: 1358102542.875
configs:
- config_name: default
data_files:
- split: train
path: data/train-*