geoskyr/mmcd-StackQA
收藏资源简介:
这是一个包含图像和对话的多模态数据集,用于训练或评估对话生成模型。数据集包含20个训练样本,每个样本包括一个图像字段和两个对话字段:conversations(原始对话列表,其中每个对话条目有from(来源)和value(内容)属性)和conversations_english(英语对话列表,其中每个对话条目有content(内容)和role(角色)属性)。数据集总大小约为772KB,适用于自然语言处理和计算机视觉的交叉任务。
This is a multimodal dataset containing images and conversations, designed for training or evaluating dialogue generation models. The dataset includes 20 training examples, each consisting of an image field and two conversation fields: conversations (a list of original dialogues with each entry having from (source) and value (content) attributes) and conversations_english (a list of English dialogues with each entry having content (content) and role (role) attributes). The total dataset size is approximately 772KB, suitable for cross-domain tasks in natural language processing and computer vision.




