Search papers, labs, and topics across Lattice.
4
0
7
Current LLMs often overfit and fail to leverage past knowledge, with naive in-context learning outperforming dedicated memory systems in continual learning tasks.
Stateful Visual Encoders enable VLMs to leverage prior visual context, leading to substantial performance gains in multi-image tasks.
Stop generating reasoning traces that merely justify actions; instead, generate traces that actually *predict* them, and watch your user modeling performance soar.
Humans are softies: AI agents can learn to win more by being more aggressive in negotiations, outperforming human players in a mixed-motive game.
