pm-25/clembench-rlvr-dataset-full-patched
收藏Hugging Face2025-09-22 更新2025-11-01 收录
下载链接：
https://hf-mirror.com/datasets/pm-25/clembench-rlvr-dataset-full-patched
下载链接
链接失效反馈官方服务：
资源简介：
这是一个包含游戏和对话数据的数据集，适用于Meta Llama 3.1模型。数据集包含id、query、reward、game、episode、origin、player（可选）和format等字段，其中query字段是从SFT-Final-Dataset和DPO-Dialogue派生出来的。数据集以Parquet文件格式存储，并提供了train split，共有27616个示例。
This dataset contains game and dialogue data, suitable for the Meta Llama 3.1 model. It includes fields such as id, query, reward, game, episode, origin, player (optional), and format, with the query field derived from SFT-Final-Dataset and DPO-Dialogue. The dataset is stored in the Parquet file format and provides a train split with a total of 27,616 examples.
提供机构：
pm-25
