Search papers, labs, and topics across Lattice.
1
0
3
Forget incomplete video descriptions: a new dataset and training pipeline yields video captioning models that are competitive with Gemini-3-Pro while reducing hallucinations.
