metadata
dataset_info:
features:
- name: text
dtype: string
- name: decode_text
dtype: string
- name: WER
dtype: float64
- name: CER
dtype: float64
- name: semantic_token
dtype: string
splits:
- name: train
num_bytes: 969021
num_examples: 4
- name: validation
num_bytes: 240149
num_examples: 1
download_size: 286203
dataset_size: 1209170
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*