metrum-ai/manipulation-resistant-prompts-1536-96

该数据集包含了具有指定目标词数的提示语，用于测试和评估满足不同长度要求的语言模型。提示语由自然语言文本组成，具有连贯的语义内容，不是合成token序列或随机字符字符串。该数据集使用词数作为统一的标准，以反映现实世界的使用模式，并避免性能测量的误导。数据集共有493个提示语，创建于2025年11月8日，目标提示语长度为1536词，目标输出长度为96词。

This dataset contains prompts with specified target word counts for both input prompts and target outputs, designed to test and evaluate language models across different length requirements. The prompts are composed of natural language text with coherent semantic content, not synthetic token sequences or random character strings. The dataset uses word counts as a universal standard to reflect real-world usage patterns and avoid misleading performance measurements. It consists of 493 prompts, created on November 8, 2025, with a target prompt length of 1536 words and a target output length of 96 words.