mlfoundations-dev/airoboros_stage_2_gpt-4o-mini_v1_3_2x

该数据集包含77257个训练样本，主要用于训练对话或问答系统。数据集包含多个字段，如指令（instruction）、响应（response）和嵌入（embedding），其中嵌入字段可能用于表示文本的向量形式。此外，数据集还包含用于检测文本相似性的字段，如too_similar和similar_text_distance，这些字段可能用于去重或相似性检测。

This dataset contains 77,257 training samples and is primarily used for training dialogue or question-answering systems. It includes multiple fields such as instruction, response, and embedding, where the embedding field likely represents text in vector form. Additionally, the dataset includes fields for detecting text similarity, such as too_similar and similar_text_distance, which may be used for deduplication or similarity detection.