New from Google: LLMs Need Sleep and Dreaming.
Researchers introduced a phase, where the model consolidates its fragile short-term memories into stable long-term memories, and then dreams to recursively self-improve over time.
For memory consolidation, researchers introduced a new form of distillation, called Knowledge Seeding (KS), where a small model(s) distills its knowledge to a larger model.
Experiments on continual learning and reasoning tasks show that this new phase can help the model to perform better and relatively better mitigates catastrophic forgetting.
Post #2642
670
Forwarded from All about AI, Web 3.0, BCI
arXiv.org Language Models Need Sleep: Learning to Self-Modify and... The past few decades have witnessed significant advances in the design of machine learning algorithms, from early studies on task-specific shallow models to more general deep Large Language Models...