carbonteq/rg-letter_counting-instruct-100k

---
license: mit
pretty_name: RLVR (verl) dataset
---

# RLVR generated dataset

Procedural rows from [reasoning-gym](https://github.com/open-thought/reasoning-gym), formatted for [verl](https://github.com/verl-project/verl) GRPO.

## Build metadata

```json
{
"config": "/home/owais/Projects/rlvr/rlvr/configs/datasets/letter_counting-instruct.yaml",
"template_type": "qwen-instruct",
"developer_prompt": null,
"data_source": "reasoning_gym",
"default_extract": "answer_tag",
"train_rows": 100000,
"test_rows": 4096,
"train_seed": 42,
"test_seed": 43,
"tasks": {
"letter_counting": {
"weight": 1,
"config": {}
}
},
"reasoning_gym_version": "0.1.19"
}
```

license: MIT许可证
pretty_name: RLVR（verl）数据集
---

# RLVR 生成数据集
本数据集包含源自[reasoning-gym](https://github.com/open-thought/reasoning-gym)的程序化生成数据行，适配[verl](https://github.com/verl-project/verl)的GRPO训练流程。

## 构建元数据
json
{
"配置文件路径": "/home/owais/Projects/rlvr/rlvr/configs/datasets/letter_counting-instruct.yaml",
"模板类型": "qwen-instruct",
"开发者提示词": null,
"数据源": "reasoning_gym",
"默认提取字段": "answer_tag",
"训练集样本行数": 100000,
"测试集样本行数": 4096,
"训练集随机种子": 42,
"测试集随机种子": 43,
"任务配置": {
"字母计数（letter_counting）": {
"任务权重": 1,
"任务专属配置": {}
}
},
"reasoning-gym版本": "0.1.19"
}