kotoba-speech/full-translation-overfit-en-ja-6-23

该数据集包含两个主要配置：feats和text。feats配置包括音频和文本的token信息，特征包括key、lang、dataset_id、spk_embed、audio_tokens_bpe、text_tokens_llama和gpt3.5_tokens_llama。text配置仅包含文本信息，特征包括key、lang、dataset_id、text_gpt3.5和text。feats配置有train和val两个划分，而text配置仅包含val划分。数据集的总下载大小为1660322字节，总数据集大小为28208304字节。

The dataset contains two main configurations: feats and text. The feats configuration includes audio and text token information, with features such as key, lang, dataset_id, spk_embed, audio_tokens_bpe, text_tokens_llama, and gpt3.5_tokens_llama. The text configuration only includes text information, with features such as key, lang, dataset_id, text_gpt3.5, and text. The feats configuration has both train and val splits, while the text configuration only has a val split. The total download size of the dataset is 1660322 bytes, and the total dataset size is 28208304 bytes.