metadata
dataset_info:
features:
- name: src
dtype: string
- name: ref
dtype: string
- name: mt
dtype: string
- name: score
dtype: float64
splits:
- name: train
num_bytes: 22489313
num_examples: 49090
download_size: 3841732
dataset_size: 22489313
configs:
- config_name: default
data_files:
- split: train
path: data/train-*