Computer SciencearXiv
Titans: Learning to Memorize at Test Time
A. Behrouz, P. Zhong, et al.
Discover Titans: a new family of architectures that pair a neural long-term memory module with attention to capture massive historical context while keeping fast, parallelizable training and inference. Experiments show Titans outperform Transformers and modern linear recurrent models on language modeling, common-sense reasoning, genomics, and time series, and can scale beyond 2M context windows. Research conducted by Ali Behrouz, Peilin Zhong, and Vahab Mirrokni.
Related Publications
Explore these studies to deepen your understanding
Adjacent work that informs or extends this paper's methodology and findings.
Physics
Deep learning at the edge enables real-time streaming ptychographic imaging
A. V. Babu, T. Zhou, et al.
Chemistry
An end-to-end deep learning framework for translating mass spectra to de-novo molecules
E. E. Litsa, V. Chenthamarakshan, et al.
Health and Fitness
Rethinking aerobic exercise intensity prescription in adults with spinal cord injury: time to end the use of "moderate to vigorous" intensity?
M. J. Hutchinson and V. L. Goosey-tolfrey
Psychology
Ageing is associated with disrupted reinforcement learning whilst learning to help others is preserved
J. Cutler, M. K. Wittmann, et al.
