Joost van de Weijer

Prompt-based adaptation can significantly enhance the merging of specialized models, yielding better performance without the pitfalls of traditional weight merging.

Forget gradient descent: this new method routes transformer activations through a Hopfield-inspired memory in a single forward pass to achieve state-of-the-art online continual learning.

Forget catastrophic forgetting: modular memory, blending in-context and in-weight learning, offers a practical path to truly continual learning agents.

Weight regularization, often overlooked in parameter-efficient continual learning, can still significantly improve the stability-plasticity trade-off, even when using low-rank adapters.