Search papers, labs, and topics across Lattice.
Affiliation:
4
0
8
6
A novel gating mechanism allows LLMs to learn new tasks without forgetting old ones, even without access to prior training data.
Supervised Memory Training enables RNNs to learn long-range dependencies more effectively while training in parallel, outperforming traditional methods.
LLMs trained with Vector Policy Optimization (VPO) learn to produce diverse solutions that unlock previously unsolvable problems in evolutionary search, outperforming models optimized for single scalar rewards.
Forget more data: pre-training on just 164M tokens of synthetic data from Neural Cellular Automata can outperform pre-training on 1.6B tokens of natural language for downstream LLM tasks.
