Search papers, labs, and topics across Lattice.
4
25
6
10
SpatialCLI enables VLMs to achieve a staggering 84.6% accuracy on spatial reasoning tasks, far surpassing existing models.
Easier tasks can sabotage the learning of harder tasks in multi-task RL, but a new entropy-aware optimization strategy can turn this challenge into an advantage.
Collective Skill Tree Search transforms LLMs into versatile agents capable of mastering complex tasks through a structured skill tree that enhances their adaptability and performance.
DPO's rise as a computationally efficient alternative to RLHF for LLM alignment has spurred a diverse range of research, now systematically organized and analyzed in this comprehensive survey.
