USS-Inferprise/Slopasaurus-Training-Slop-Stabilized
收藏Hugging Face2026-04-24 更新2026-04-26 收录
下载链接：
https://hf-mirror.com/datasets/USS-Inferprise/Slopasaurus-Training-Slop-Stabilized
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是经过修改的版本，基于USS-Inferprise/Slopasaurus-Training-Slop和yahma/alpaca-cleaned数据集。修改包括：注入|SLOP|作为激活词以创建slop内容；替换原型名称以防止过度训练原型标记；并添加了1000行来自yahma/alpaca-cleaned的常规数据。该数据集最终用于训练Slopasaurus模型。
This modified dataset using USS-Inferprise/Slopasaurus-Training-Slop and yahma/alpaca-cleaned. Methodology includes: Inject |SLOP| as activation word for slop creation; Replace the archetype names with personal names to prevent overtraining of the archetype tokens; Added 1000 conventional rows from yahma/alpaca-cleaned. This repo is the resulting training set that we used to train Slopasaurus.
提供机构：
USS-Inferprise
