dddraxxx/sealvqa-vaw-spatial-10k-hardsamples

这是一个包含了图像和文本数据的数据集，具有多个字段，包括但不限于图像字段(images)，数据来源(data_source)，奖励模型(reward_model)，以及额外信息(extra_info)。额外信息中包含了答案(answer)，边界框格式(bbox_format)，边界框格式描述(bbox_format_description)，会话轮数(conversation_turns)，数据分割(data_split)，数据集类型(dataset_type)，是否包含目标(has_targets)，图像路径(image_path)，索引(index)，原始对话(original_conversations)，问题(question)，问题ID(question_id)，搜索(search)，目标实例(target_instances)。此外，数据集还提供了prompt字段，其中包含内容和角色信息。数据集还提供了ability和env_name字段。数据集分为训练集，其大小为888422546字节，包含2068个示例。

This dataset contains images and text data with multiple fields including but not limited to images, data_source, reward_model, and extra_info. The extra_info includes answer, bbox_format, bbox_format_description, conversation_turns, data_split, dataset_type, has_targets, image_path, index, original_conversations, question, question_id, search, and target_instances. Additionally, the dataset provides a prompt field containing content and role information. It also includes ability and env_name fields. The dataset is split into a training set, which is 888422546 bytes in size and contains 2068 examples.