meituan-longcat/UNO-Scorer-Qwen3-14B

UNO-Scorer-Qwen3-14B by meituan-longcat is a 14 billion parameter LLM-based evaluation model built on the Qwen3-14B architecture, designed for automated scoring of Large Multimodal Models (LMMs). Fine-tuned on 13K in-house data, it provides numerical scores and detailed reasoning by comparing sub-questions against reference answers. This model excels in evaluating complex Multi-Step Open-Ended Questions, demonstrating superior accuracy compared to models like GPT-4.1 in this specific domain.