Search papers, labs, and topics across Lattice.
1
0
2
High artifact detection rates in VLMs mask significant failures in contextual understanding, with top models misidentifying visual cues in over 46% of cases.
