quablab/details_quablab__smollm3-dpo-aligned_private
收藏Hugging Face2025-10-15 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/quablab/details_quablab__smollm3-dpo-aligned_private
下载链接
链接失效反馈官方服务：
资源简介：
在模型quablab/smollm3-dpo-aligned评估过程中自动创建的数据集，包含一个配置，每个配置对应一个评估任务。数据集由一次运行创建，每次运行在配置中作为特定分割存在，使用运行时间戳命名。train分割指向最新结果，results配置存储所有聚合结果。
Dataset automatically created during the evaluation of model quablab/smollm3-dpo-aligned, containing one configuration corresponding to one evaluation task each. The dataset is created from one run, with each run represented as a specific split in the configuration named using the runs timestamp. The train split always points to the latest results, and the results configuration stores all aggregated results.
提供机构：
quablab
