ChatDoctor-RL

# Intelligent-Internet/ChatDoctor-Improved-Answer Dataset

This dataset represents a carefully curated subset derived from the original ChatDoctor-HealthCareMagic-100k[lavita/ChatDoctor-HealthCareMagic-100k] dataset, where we have undertaken significant improvements to enhance the quality and depth of the responses. The answers have been thoroughly refined to provide greater detail, clarity, and precision, while incorporating a heightened focus on safety awareness to ensure responsible and secure usage.

# Data Decontamination

To ensure the integrity and reliability of the dataset for RL training, a rigorous two-step decontamination process was applied:

## 8-grams Decontamination

Followed the open-r1 methodology to identify and eliminate overlap with evaluation datasets using 8-gram sequences.

This step ensures that the dataset does not contain sequences that could bias evaluation results.

## Fuzzy Decontamination

Applied the s1k method with a stringent 80% similarity threshold to further remove any near-duplicate or highly similar samples.

This additional step guarantees minimal overlap with evaluation datasets, preserving the dataset's purity.

# Intelligent-Internet/ChatDoctor-Improved-Answer 数据集

本数据集源自原始ChatDoctor-HealthCareMagic-100k[lavita/ChatDoctor-HealthCareMagic-100k]数据集的精心筛选子集，我们对其开展了大幅优化工作，以提升回复内容的质量与深度。所有答案均经过全面精修，以提供更丰富的细节、更清晰的逻辑与更精准的表述，同时进一步强化了安全意识相关内容，确保数据集可被负责任且安全地使用。

# 数据净化（Data Decontamination）

为确保用于强化学习（RL）训练的数据集的完整性与可靠性，我们采用了严格的两步数据净化流程：

## 8-gram 数据净化（8-grams Decontamination）

本次流程遵循open-r1方法学，通过8元组（8-gram）序列识别并移除与评测数据集存在重叠的内容。该步骤可确保数据集不存在可能导致评测结果出现偏差的序列片段。

## 模糊净化（Fuzzy Decontamination）

本次流程采用s1k方法，设置严格的80%相似度阈值，进一步移除所有近似重复或高度相似的样本。该额外步骤可确保数据集与评测数据集的重叠度极低，保障数据集的纯净性。