five

rl-rag-2/oss-high-drtulu-v5-0422-8rollouts

收藏
Hugging Face2026-04-28 更新2026-05-03 收录
下载链接:
https://hf-mirror.com/datasets/rl-rag-2/oss-high-drtulu-v5-0422-8rollouts
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集名为OSS High-Effort Trajectories: drtulu_v2_bc_synthetic_v5_0422,是一个正在进行中的高推理努力轨迹数据集,基于`gpt-oss-120b`模型生成。数据集源自`rl-rag-2/drtulu_v2_bc_synthetic_v5_0422`的train分割,包含4,129个问题(当前已完成2375个问题的数据),每个问题有8次rollout,总计已完成18980/33032条轨迹。每条轨迹数据以JSONL格式存储,包含问题ID、问题文本、参考回答、轨迹索引、完整对话记录、搜索URL列表、搜索结果记录、生成时间及状态等信息。支持的工具包括浏览器搜索、页面打开和文本查找功能。

The dataset named OSS High-Effort Trajectories: drtulu_v2_bc_synthetic_v5_0422 is an in-progress high-reasoning-effort trajectory dataset generated by the `gpt-oss-120b` model. It is derived from the train split of `rl-rag-2/drtulu_v2_bc_synthetic_v5_0422`, containing 4,129 questions (currently with data for 2,375 questions) and 8 rollouts per question, totaling 18,980/33,032 completed trajectories. Each trajectory is stored in JSONL format, including question ID, question text, reference answer, trajectory index, full conversation records, searched URLs list, search results records, generation time, and status. Supported tools include browser search, page opening, and text finding functions.
提供机构:
rl-rag-2
5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作