ZTE-AIM/32B_LLM_AdaptiveCode_data

该数据集是基于LLM-Adaptive方法构建的，用于训练细粒度推理模型，特别针对数学问题求解和代码生成任务。数据集通过动态构建与目标LLM问题解决能力相匹配的难度对齐的数据集来实现模型的训练。

The dataset is built based on the LLM-Adaptive approach for training fine-grained reasoning models, particularly for mathematical problem solving and code generation tasks. The dataset enables model training by dynamically constructing difficulty-aligned datasets that match the problem-solving capabilities of the target LLM.