zilliz/mmarco_chinese_200k-context-relevance-with-think

该数据集包含带有上下文跨度相关性注释的查询-上下文对，用于训练RAG系统中的语义高亮模型。关键特征包括context_spans（指示上下文文本中分段跨度的位置）、context_spans_relevance（二进制标签，指示每个跨度是否应被高亮显示）和think_process（包含注释过程中使用的推理过程，以提高标签的准确性和可解释性）。数据集来源于unicamp-dl/mmarco的一个子集，并使用Qwen3-8B进行注释。

This dataset contains query-context pairs with relevance annotations for context spans. The annotations help identify which parts of a document are semantically relevant to a query, even when they dont contain exact keyword matches. Key features include context_spans (indicates the positions of segmented spans within the context text), context_spans_relevance (binary labels indicating whether each span should be highlighted), and think_process (contains the reasoning process used during annotation to ensure more accurate context span relevance labels and improve observability and interpretability). The dataset was annotated using Qwen3-8B and is derived from a subset of unicamp-dl/mmarco.