Search papers, labs, and topics across Lattice.
4
0
6
12
Captions can be made 48% more complete and hallucination reduced by 45% without retraining models, thanks to a novel prominence-aware rectification approach.
BackgroundMellow achieves unprecedented narrative cohesion in audio generation, transforming how we create immersive soundscapes for storytelling.
Direction-Conditioned Policies achieve substantial performance improvements in goal-conditioned reinforcement learning by effectively utilizing gradient information, outperforming existing methods in complex environments.
Escape the CVAE decoder bottleneck: SPAARS unlocks better offline-to-online RL by safely exploring a latent space, then seamlessly switching to raw actions.
