Search papers, labs, and topics across Lattice.
Indian Institute of Technology Kharagpur
1
0
3
VLMs hallucinate less when you force them to "think twice" by contrasting language-driven and vision-driven token probabilities at each decoding step.
