Multimodal large language models (MLLMs) are rapidly advancing clinical diagnosis, yet their adaptation pipelines remain anchored to accuracy-based objectives. Clinical data are heavily class-imbalanced: a constant-majority predictor can score above 90% accuracy while being clinically useless. We therefore evaluate and...
This work introduces DEO, which replaces challenger parameter updates with solver-guided task sampling, and shows that DEO learns distributionally robust reasoning ability.
A pairwise validator is proposed: a frozen LLM that, given the parent and child candidate, returns a binary verdict on which is better, which mitigates the need for strict scale calibration.
Minghao Liu, Yu Wang, Jiayun Wang et al.· arXiv.org· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.