MM Biscuit V2
MM Biscuit V2 is a multimodal AI evaluation project focused on improving how AI models understand and respond to both text and image based inputs. The project involves reviewing user prompts, analysing attached images where applicable, evaluating AI generated responses, and checking whether the model follows instructions accurately. It also includes identifying errors, rating response quality, assessing image-text alignment, and writing clear justifications for the chosen labels or rankings.