arXiv:2606.09248v1 Announce Type: new 
Abstract: Multi-modal Large Language Models (MLLMs) have achieved remarkable progress in video temporal grounding with reinforcement learning for generating reasoning paths. However, existing models often produce superficial reasoning, which offers limited guidance for precise temporal localization. This limitation stems from (1) inefficient random exploration and (2) reward functions that focus solely on the answer correctness while ignoring reasoning quality. To address these issues, we propose TaRO (Temporal-Aware Reasoning Optimization), a framework that explicitly enhances the model's ability of thinking with time. First, we introduce a Constructive Reasoning Exploration that leverages pre-generated dense captions to construct reasoning paths grounded in explicit visual cues and timestamps, enabling efficient exploration of high-quality time-aware reasoning. Second, to evaluate reasoning quality, we design a Temporal-Sensitivity Reward. High-quality reasoning should be anchored to specific events and timestamps. If the event boundary under thinking is disrupted, such reasoning should become invalid, leading to a drop in the logit of the reasoning path. We utilize this drop as a critique of reasoning quality. Finally, TaRO follows a progressive curriculum, which starts by utilizing this reward to select better constructed reasoning paths, and evolves to a free exploration phase where the model autonomously generates effective reasoning. Experiments demonstrate that TaRO achieves state-of-the-art performance on VTG benchmarks. Code is available at https://github.com/oceanflowlab/TaRO.

يمثل إدخال تحسينات في التفكير الزمني (TaRO) تقدمًا كبيرًا في تحديد المواقع الزمنية للفيديو من خلال تعزيز قدرات التفكير لنماذج اللغة متعددة الوسائط (MLLMs). يتناول هذا الإطار القيود المفروضة على النماذج الحالية، التي غالبًا ما تنتج تفكيرًا سطحيًا، من خلال دمج استكشاف تفكير بنّاء ونظام مكافأة حساسية زمنية.

La introducción de la Optimización de Razonamiento Sensible al Tiempo (TaRO) marca un avance significativo en la localización temporal de videos al mejorar las capacidades de razonamiento de los Modelos de Lenguaje Multimodal (MLLMs). Este marco aborda las limitaciones de los modelos existentes, que a menudo producen razonamientos superficiales, al incorporar una Exploración de Razonamiento Constructivo y un sistema de Recompensa de Sensibilidad Temporal.

L'introduction de l'Optimisation de Raisonnement Sensible au Temps (TaRO) représente une avancée significative dans le domaine de la localisation temporelle des vidéos en améliorant les capacités de raisonnement des Modèles de Langage Multi-modal (MLLMs). Ce cadre s'attaque aux limitations des modèles existants, qui produisent souvent un raisonnement superficiel, en intégrant une Exploration de Raisonnement Constructif et un système de Récompense de Sensibilité Temporelle.

The introduction of Temporal-Aware Reasoning Optimization (TaRO) marks a significant advancement in video temporal grounding by enhancing the reasoning capabilities of Multi-modal Large Language Models (MLLMs). This framework addresses limitations in existing models, which often produce superficial reasoning, by incorporating a Constructive Reasoning Exploration and a Temporal-Sensitivity Reward system.