gss1147/god_level_producer_mindframe_dataset

这个名为God-Level Producer Mindframe Dataset (15k)的数据集旨在训练大型语言模型（LLMs）像传奇的嘻哈和节奏制作制作人一样思考和制作音乐。它包含15,000个示例，涵盖了音乐制作的各个方面，如声音设计、编曲、混音、母带处理和重新混音。数据集还详细说明了特定乐器、插件和数字音频工作站（DAWs）的使用，以及制作决策背后的原因。数据集的结构包括id、category、difficulty、producer_style、daw、instruction、input、output和tags等字段。平均输出长度在1,800到2,200个字符之间。README还提供了如何加载数据集的示例，并推荐了LLMs的微调方法。

The God-Level Producer Mindframe Dataset (15k) is designed to train large language models (LLMs) to think and produce music like legendary hip-hop and beat-making producers. It includes 15,000 examples covering various aspects of music production such as sound design, arrangement, mixing, mastering, and remixing. The dataset also details the use of specific instruments, plugins, and digital audio workstations (DAWs), along with the reasoning behind production decisions. The structure of the dataset is outlined with fields like id, category, difficulty, producer_style, daw, instruction, input, output, and tags. The average output length is between 1,800 to 2,200 characters. The README also provides an example of how to load the dataset and recommends fine-tuning methods for LLMs.