🎉 Paper “Kalman Linear Attention: Parallel Bayesian Filters For Efficient Language Modelling” accepted at the International Conference on Machine Learning 2026. Joint work with Cameron, Aidan, Andras, Elliot and Amos. arXiv