kimi-k-set
收藏资源简介:
Kimi-K-Set Reasoning Dataset 是一个用于训练推理模型(CoT)的组合数据集,包含蒸馏数据和监督微调(SFT)数据。数据集结构包括三个主要字段:`instruction`(输入请求)、`thought`(推理链,Chain-of-Thought)和 `output`(最终答案)。数据集分为训练集和测试集,训练集由多个 JSONL 文件组成,测试集为一个单独的 JSONL 文件。数据集采用 Apache-2.0 许可证。
The Kimi-K-Set Reasoning Dataset is a composite dataset for training Chain-of-Thought (CoT) based reasoning models, which contains distillation data and Supervised Fine-Tuning (SFT) data. The dataset structure includes three core fields: `instruction` (input request), `thought` (Chain-of-Thought reasoning chain), and `output` (final answer). The dataset is divided into training and test sets: the training set consists of multiple JSONL files, while the test set is a single standalone JSONL file. The dataset is licensed under Apache-2.0.




