|
- GitHub - decoderesearch SAELens: Training Sparse Autoencoders on . . .
Migrating to SAELens v6 The new v6 update is a major refactor to SAELens and changes the way training code is structured Check out the migration guide for more details
- SAELens:用于语言模型的稀疏自编码器训练与分析工具 - 懂AI
SAELens是一个专门用于训练和分析语言模型中的稀疏自编码器的开源工具库,旨在帮助研究人员深入理解神经网络内部机制,为创建安全可靠的AI系统提供洞见。
- sae-lens · PyPI
Training and Analyzing Sparse Autoencoders (SAEs)
- Releases · decoderesearch SAELens - GitHub
Training Sparse Autoencoders on Language Models Contribute to decoderesearch SAELens development by creating an account on GitHub
- SAELens - 训练和分析稀疏自编码器的开源工具 - 懂AI
SAELens是一个开源工具库，专注于稀疏自编码器的训练和分析。 它为研究人员提供预训练模型加载、自定义训练和可视化分析功能，支持深入探索神经网络内部机制。 该项目由多位贡献者维护，旨在促进机械解释性研究和人工智能安全发展。
- GitHub - PKU-Alignment SAELens-V
Building on SAELens, we developed SAE-V to facilitate training multi-modal models, such as LLaVA-NeXT and Chameleon Additionally, we created a series of scripts that use SAE-V to support mechanistic interpretability analysis in multi-modal models
- SAELens tutorials tutorial_2_0. ipynb at main - GitHub
However, we will explain what SAE features are, how to load SAEs into SAELens and find identify features, and how to do steering, ablation, and attribution with them
- Wouter Saelens - Google Scholar
Wouter Saelens Professor in perturbational systems biology Expert Scientist at VIB Inflammation Research Center Verified email at ugent be Computational biology
|
|
|