TLDR Tech / AI
An off switch for dual use knowledge in AI models
GRAM (Gradient-Routed Auxiliary Modules) enables the benefits of training many separately-filtered models at the cost of training only one model. It gives models dedicated, removable compartments for each category of dual-use knowledge and only updates those compartments when learning from dual-use data. It allows parts of the model's knowledge to be deleted after training, providing a stronger framework for model safety.
Read the full story