SpeechPPL/SALMon_Flow-SLM_270m

该数据集用于音频处理任务，包括不同的配置，每个配置都有特定的特征和数据分割。数据集似乎用于训练模型执行音频延续任务，其中模型被训练以生成给定音频提示的延续。数据集包含各种特征，如任务类型、索引、正负音频样本、提示音频、延续音频、标记级损失值、模型生成的延续，以及与音频和模型配置相关的各种其他元数据。

This dataset is used for audio processing tasks, including different configurations, each with specific features and data splits. The dataset appears to be used for training models on audio continuation tasks, where models are trained to generate continuations of given audio prompts. The dataset contains various features such as task type, indices, positive and negative audio samples, prompt audio, continuation audio, token-wise loss values, model-generated continuations, and various other metadata related to the audio and model configurations.