Search papers, labs, and topics across Lattice.
Stevens Institute of Technology
1
0
2
Unfaithful reasoning chains can significantly undermine the accuracy of MLLMs, but a training-free framework can effectively correct these errors and boost performance by over 8%.
