Search papers, labs, and topics across Lattice.
ByteDance
1
0
2
Reinforcement learning enables video-LLMs to re-watch and refine answers without the costly overhead of chain-of-thought training, achieving better performance with less computation.
