inclusionAI/Ling-Coder-SyntheticQA

Ling-Coder数据集包括以下部分：Ling-Coder-SFT（用于训练Ling-Coder Lite的SFT数据子集，包含超过500万个样本）、Ling-Coder-DPO（用于训练Ling-Coder Lite的DPO数据子集，包含25万个样本）和Ling-Coder-SyntheticQA（用于Ling-Coder Lite退火训练的合成数据子集，包含超过2400万个英文和中文样本）。这些数据覆盖了文本到代码、文本到SQL、代码执行推理、代码修复和测试用例生成等多种任务主题。数据集是通过Magpie方法使用开源代码LLMs如Qwen2.5-Coder和DeepSeek-Coder-v2生成的，并经过了规则清洗、脱毒、去重、质量检查和消融测试的流程。

The Ling-Coder Dataset includes the following components: Ling-Coder-SFT (a subset of SFT data used for training Ling-Coder Lite, containing more than 5 million samples), Ling-Coder-DPO (a subset of DPO data used for training Ling-Coder Lite, containing 250k samples), and Ling-Coder-SyntheticQA (a subset of synthetic data used for annealing training of Ling-Coder Lite, containing over 24 million English and Chinese samples). The dataset covers a variety of task topics such as text2code, text2sql, code execution reasoning, code fixing, and testcase generation. The dataset was generated using the Magpie method with open-source code LLMs like Qwen2.5-Coder and DeepSeek-Coder-v2, and underwent a pipeline of rule-based cleansing, detoxification, deduplication, quality inspection, and ablation testing to meet our admission criteria.