mlfoundations-dev/instruction_filtering_scale_up_code_base_gemini_length

该数据集包含多个字段，用于记录指令种子(instruction_seed)、来源(source)、Gemini模型响应(gemini_response)、原始行索引(__original_row_idx)、长度(length)、推理过程(reasoning)、DeepSeek解决方案(deepseek_solution)以及对话(conversations)。数据集仅包含训练集(train)，共有16000个示例，总大小为753546514字节。

The dataset includes multiple fields for instruction seed (instruction_seed), source (source), Gemini model response (gemini_response), original row index (__original_row_idx), length (length), reasoning process (reasoning), DeepSeek solution (deepseek_solution), and conversation (conversations). The dataset contains only the training set (train) with a total of 16,000 examples and a total size of 753,546,514 bytes.