saklas is an open-source tool for activation steering and trait monitoring in HuggingFace transformer models. It aids AI researchers and ML engineers in understanding and controlling model behavior, supporting mechanistic interpretability and probing of large language models.
saklas sits in PulseGate's LLM eval & observability category. It focuses on providing tools for interpreting and steering activations in transformer-based language models. It is built as an open-source project for AI researchers and ML engineers. saklas is open source under the AGPL-3.0 license. It runs on the web and the command line.
a9lim builds and maintains saklas, and the product first shipped in 2026. Development happens publicly on GitHub with 765 commits in the last 90 days. Key capabilities include activation steering, trait monitoring, and transformer support.
Latest indexed changes and source events
Other apps tracked under the same category.
