jimnoneill/BSG_CyLlama-computational-training

BSG CyLlama V8 - 计算训练数据是一个包含20,000个计算主题训练样本的数据集，由DeepSeek生成。每个样本包含来自科学聚类的源摘要以及目标输出（如摘要摘要、简短摘要、标题、概述），用于训练BSG CyLlama聚类描述生成器。数据格式为TSV，列包括cluster_id、theme、source_abstracts、abstract_summary、short_summary、title、theme_token、short_title和overview。BSG CyLlama V8 Computational Training Dataset is a dataset containing 20,000 training samples focused on computational topics, generated by DeepSeek. Each sample includes source abstracts from scientific clusters and target outputs such as abstract summaries, short summaries, titles, and overviews, which are used to train the BSG CyLlama cluster description generator. The dataset is stored in TSV format, with columns including cluster_id, theme, source_abstracts, abstract_summary, short_summary, title, theme_token, short_title, and overview.