---
title: II-Medical-RL
canonical_url: "https://www.modelscope.cn/datasets/Intelligent-Internet/II-Medical-RL"
md_url: "https://www.modelscope.cn/datasets/Intelligent-Internet/II-Medical-RL.md"
repository: Intelligent-Internet/II-Medical-RL
last_updated: 2025-07-04
license: "Apache License 2.0"
storage_size: "24 MB"
downloads: 363
stars: 0
---

# II-Medical-RL

> II-Medical-RL - Intelligent-Internet 在 ModelScope 开源的数据集。The MedReason-RL dataset is a refined version of the original MedReason dataset, specifically curated for training reinforcement learning (RL) models to enhance reasoning abilities. It has been proven…

Intelligent-Internet/II-Medical-RL 是 ModelScope 魔搭社区上的数据集，存储大小 24 MB，采用 Apache License 2.0 许可。

- **Repository**: Intelligent-Internet/II-Medical-RL
- **License**: Apache License 2.0
- **Storage size**: 24 MB
- **Downloads**: 363
- **Stars**: 0
- **Last updated**: 2025-07-04

Source: https://www.modelscope.cn/datasets/Intelligent-Internet/II-Medical-RL

---

# Overview

The MedReason-RL dataset is a refined version of the original [MedReason dataset](https://huggingface.co/datasets/UCSC-VLAA/MedReason), specifically curated for training reinforcement learning (RL) models to enhance reasoning abilities. It has been proven to be the best dataset for improving model reasoning through RL training.

# Source

This dataset is derived from the original MedReason dataset, which focuses on medical reasoning tasks. However, the original dataset contained significant overlap with benchmarking datasets, necessitating decontamination.

# 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.

# Conclusion

The decontamination process ensures that the MedReason-RL dataset is free from contamination by benchmarking datasets, making it a reliable and high-quality resource for training RL models to improve reasoning abilities in medical contexts.