mlfoundations-dev/opencoder

这是一个包含对话数据的训练集，数据集包含对话的起始指令（instruction_seed）、预期的响应（response_seed）、实际响应（response）以及原始数据行索引（__original_row_idx）。此外，每个样本还包含一个对话列表（conversations），其中包含对话的来源（from）和内容（value）。训练集包含50000个示例。This is a training dataset containing dialogue information, which includes the starting instruction of the dialogue (instruction_seed), the expected response (response_seed), the actual response (response), and the index of the original data row (__original_row_idx). In addition, each sample contains a list of conversations, including the source (from) and content (value) of the conversation. The training set contains 50,000 examples.