We have been calling this the 'reasoning' / 'non-reasoning' split, but this is not a perfectly clean dichotomy. Several correlated but not strictly identical changes happened over the same few months: scaling inference compute, heavier use of RL in post-training, and models producing reasoning tokens.
这里承认了分类方法的局限性,指出2024年左右的AI能力加速可能是由多个因素共同作用的结果,而非仅仅是推理能力的提升。这表明文章作者对数据的复杂性有清醒认识,但缺乏对这些因素相对重要性的量化分析。