
J-Wash
Extraltodeus
Jacobian-Brainwash : A framework built on Anthropic's Jacobian Lens for analyzing and customizing LLM internal representations with exportable results.
AI 简介
J-Wash 是一个基于 Anthropic Jacobian Lens 的本地化 LLM 表征编辑框架,支持在不训练、不依赖数据集的前提下,通过修改 token 级别方向来重塑模型行为,并导出为标准 safetensors 格式检查点。其核心功能包括实时 Jacobian 热力图可视化、概念级 token 方向标注与编辑、多粒度导出(全量权重/指定层/LoRA)。技术上融合了可解释性分析(Jacobian lens)、激活工程与模型权重编辑,提供 FastAPI+React 交互式界面。适用于 LLM 可解释性研究、轻量对齐实验、行为定制(如角色设定重写)及教学演示等无需微调的快速干预场景。
Python
Apache License 2.0180
Stars
17
Forks
1
Watchers
1
Issues
Star 增长
今日0
近 7 天+58
近 30 天+58
综合评分69.57
默认分支master