Search papers, labs, and topics across Lattice.
South China University of Technology
1
0
3
Even state-of-the-art LMMs struggle with reasoning-informed visual editing, with the best model only achieving 56.6% accuracy on a new comprehensive benchmark.
