Search papers, labs, and topics across Lattice.
3
0
6
Reinforcement learning can efficiently merge heterogeneous language models in federated ASR, outperforming genetic algorithms and improving character error rate.
GRPO-based fine-tuning, guided by LLM-generated captions, significantly boosts text-to-audio synthesis fidelity and prompt adherence.
Sparse-view 3D Gaussian Splatting gets a major boost by incorporating priors from geometry foundation models and VLMs to overcome the limitations of color residual heuristics.
