V-Zero：結合對比證據門控的無答案標籤同策略蒸餾用於細粒度視覺推理

摘要

細粒度視覺推理要求多模態大型語言模型（MLLMs）能辨識與任務相關的視覺證據，並將其推理過程植基於局部影像區域。現有的基於代理的方法通常依賴帶有可驗證獎勵的強化學習，或是在大規模標註推理軌跡上進行監督式微調，導致探索成本高昂、需人工設計驗證規則，或高度依賴文本監督。一個自然避開此類外部答案標籤的方式，是讓模型從自身取樣的軌跡中學習，這便導向「在策略蒸餾」（OPD）。為了理解OPD在視覺推理中能與不能提供什麼，我們將其重新審視為「無負樣本的梯度停止對齊」。此觀點顯示，儘管OPD提供了有效的詞元層級修正，其天花板卻受制於缺乏軌跡層級的區辨能力。受此觀察啟發，我們提出V-Zero，一個無需答案標籤的視覺推理框架，並採用對比證據門控機制。V-Zero不使用任何標註文本答案；相反地，在訓練過程中，它將與問題相關的區域裁剪與負視覺視角配對，用以評估學生取樣的軌跡，並門控密集詞元層級的蒸餾。在多個視覺推理基準上的實驗顯示，V-Zero持續改善細粒度視覺推理，同時保留強大的泛化能力。值得注意的是，V-Zero比先前的監督式微調方法快超過5倍，比強化學習基線快超過10倍。程式碼與資料集將於 https://github.com/eVI-group-SCU/V-Zero 釋出。

English

Fine-grained visual reasoning requires multimodal large language models (MLLMs) to identify task-relevant visual evidence and ground their reasoning in local image regions. Existing agentic methods typically rely on reinforcement learning with verifiable rewards or supervised fine-tuning on large-scale annotated reasoning traces, leading to costly exploration, hand-designed verification rules, or heavy dependence on textual supervision. A natural way to avoid such external answer labels is to learn from trajectories sampled by the student itself, which points to On-Policy Distillation (OPD). To understand what OPD can and cannot provide for visual reasoning, we revisit it as negative-free stop-gradient alignment. This perspective shows that, although OPD provides effective token-level correction, its ceiling is constrained by the absence of trajectory-level discrimination. Motivated by these observations, we propose V-Zero, an answer-label-free framework for visual reasoning with contrastive evidence gating. V-Zero uses no annotated textual answer labels; instead, during training it pairs a question-relevant regional crop with a negative visual view to evaluate student-sampled trajectories and gate dense token-level distillation. Experiments on multiple visual reasoning benchmarks show that V-Zero consistently improves fine-grained visual reasoning while preserving strong generalization. Notably, V-Zero is more than 5times faster than previous supervised fine-tuning methods and more than 10times faster than reinforcement learning baselines. Code and dataset will be released at https://github.com/eVI-group-SCU/V-Zero