Latest AI updates

MAEMs: Scaling Activation Inversion to Interpret Residual Stream Directions

Researchers introduce Maximally-Activating Example Metamodels (MAEMs), which generate text exemplars that strongly activate specific residual stream directions in a subject language model. Using a label-free reinforcement learning objective, MAEMs scale to diverse sources of directions and outperform corpus search and natural-language autoencoders on held-out activations. The approach also recovers LoRA backdoor payloads and verbalizes model inner monologue, with exemplars verifiable via forward passes.