ramen-noodels/audio_red_switch_train_unnormalized

该数据集是一个用于机器学习任务的数据集，包含训练集分割，共有15270个样本。每个样本由input_values（一个浮点数列表）和label（一个int64类型的标签）组成。数据集总大小约为2.35 GB，适用于音频处理、信号分析或分类任务，其中input_values可能表示音频特征（如梅尔频谱图），label表示对应的类别或目标值。

This dataset is designed for machine learning tasks, containing a training split with 15,270 samples. Each sample consists of input_values (a list of float32 values) and label (an int64 label). The total dataset size is approximately 2.35 GB, suitable for audio processing, signal analysis, or classification tasks, where input_values may represent audio features (e.g., mel-spectrograms) and label indicates the corresponding category or target value.