Dyno Lab — an AI safety, alignment and interpretability research workbench for Apple Silicon. Explore activations, probes, SAEs and interventions with local inference, a Python SDK, APIs and MCP.
↑ +1 today
Discover the best Mechanistic Interpretability open source repositories on GitHub. Explore trending tools, live star counts, source code, and developer projects updated daily.