Parikshith/grow-1-monolingual-1m-ha-en-scored

--- dataset_info: features: - name: src dtype: string - name: mt dtype: string - name: score_wmt20-comet-qe-da dtype: float64 - name: score_wmt21-comet-qe-da dtype: float64 - name: score_wmt23-cometkiwi-da-xl dtype: float64 splits: - name: train num_bytes: 291625402 num_examples: 1100000 download_size: 197067897 dataset_size: 291625402 configs: - config_name: default data_files: - split: train path: data/train-* ---

原始信息汇总

数据集信息

特征

• src: 类型为字符串
• mt: 类型为字符串
• score_wmt20-comet-qe-da: 类型为浮点数
• score_wmt21-comet-qe-da: 类型为浮点数
• score_wmt23-cometkiwi-da-xl: 类型为浮点数

数据分割

• train: 包含1,100,000个样本，总字节数为291,625,402

数据集大小

• 下载大小: 197,067,897字节
• 数据集大小: 291,625,402字节

配置

• default: 包含训练数据文件，路径为data/train-*