DeepMind has released Gemma Scope 2, a set of open interpretability tools for language models. These tools allow researchers to better understand complex behavior in language models, such as those in the Gemma 3 family. The release aims to aid the AI safety community's efforts to develop more transparent and explainable AI systems. --- Why it matters: This matters because it provides researchers with a deeper understanding of how language models work, which is crucial for developing safer and more reliable AI systems. Source: https://deepmind.google/blog/gemma-scope-2-helping-the-ai-safety-community-deepen-understanding-of-complex-language-model-behavior/