metadata
dataset_info:
features:
- name: ground_truth
dtype: string
- name: pixel_values
sequence:
sequence:
sequence: float32
- name: labels
sequence: int64
- name: target_sequence
dtype: string
splits:
- name: train
num_bytes: 8419528948
num_examples: 1013
- name: validation
num_bytes: 2111092299
num_examples: 254
- name: test
num_bytes: 3923067240
num_examples: 472
download_size: 2126351712
dataset_size: 14453688487
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*