jan-hq/Pick-Place-Table-Reasoning-100x100-output

该数据集包含了多个字段，如源对象(Source_Obj)、目标对象(Target_Obj)、思考过程(Thinking)、对象(Object)、指令(instruction)、解决方案(solution)和对话(Conversation)。解决方案字段由整数序列构成，对话字段则由内容(content)和角色(role)组成。数据集分为训练集(train)，包含300,000个示例，总大小约为4.4GB。提供了默认配置，指定了训练数据的路径。

The dataset includes fields such as Source_Obj, Target_Obj, Thinking, Object, instruction, solution, and Conversation. The solution field consists of an integer sequence, and the Conversation field is composed of content and role. The dataset is split into a training set (train) with 300,000 examples, totaling approximately 4.4GB in size. A default configuration is provided, specifying the path to the training data.