1 paper
Mikhail Kulakov, Ruslan Solikhov, Ivan Bondarenko
We present our system for the SHROOM-visions 2026 shared task on character-level hallucination span detection in image-conditioned vision-language model (VLM) responses, evaluated…