knowurknottty/claude-sonnet-4.6-120000x

这是一个严格专业、高级的合成数据集，旨在训练下一代模型进行高级推理、逻辑推断和多领域编程。主要教师模型是配备了自适应思维级别的Claude Sonnet 4.6，通过其动态计算分配，推理路径在快速直觉跳跃和深入多步思考之间有机转换。数据集还使用了Gemini 3.1 Pro进行交叉验证和结构复杂性处理。所有数据完全未经审查，保留了在显式、哲学和历史范围内的0拒绝率。数据集覆盖了多个领域，包括通用知识与推理（40%）、高级代码与逻辑（30%）、数学（15%）和逻辑（15%）。数据集包含90,207行通用知识与推理数据和32,166行高级代码与逻辑数据，总令牌数超过1.75亿。数据集的质量评分为9.1/10.0，具有高度多样性，涵盖了从初学者到研究生研究范式的各种难度级别。

This is a strictly professional, high-grade synthetic dataset designed to train next-generation models in advanced reasoning, logical extrapolation, and multi-domain programming. The primary teacher model driving the reasoning traces is Claude Sonnet 4.6 equipped with Adaptive Thinking Level, ensuring natural, human-like thinking styles. For cross-verification and structural complexity, Gemini 3.1 Pro was utilized. All data is fully uncensored, retaining 0 refusals across explicit, philosophical, and historical bounds. The dataset covers various domains including General Knowledge & Reasoning (40%), Advanced Code & Logic (30%), Math (15%), and Logic (15%). It contains 90,207 rows of General Knowledge & Reasoning data and 32,166 rows of Advanced Code & Logic data, with over 175 million tokens in total. The dataset has a quality score of 9.1/10.0 and guarantees global diversity, spanning from beginner introductions to post-graduate researcher paradigms.