AI News and Headlines

Google DeepMind Researchers Release Gemma Scope 2 as a Full Stack Interpretability Suite for Gemma 3 Models

Google DeepMind has unveiled Gemma Scope 2, an open‑source interpretability suite that spans the entire Gemma 3 family from 270 M to 27 B parameters. The platform leverages layer‑wise sparse autoencoders and transcoders, trained on roughly 110 petabytes of activation data and over a trillion parameters. New features include Matryoshka training for more stable concepts and dedicated tools for chat‑tuned models. Researchers can now trace jailbreaks, hallucinations, and sycophantic responses back to internal feature activations rather than only observing inputs and outputs.