Beyond Perceptual Shortcuts: Causal-Inspired Debiasing Optimization for Generalizable Video Reasoning in Lightweight MLLMs

• URL: http://arxiv.org/abs/2605.01324v2
• Date: Tue, 05 May 2026 09:38:12 GMT
• Title: Beyond Perceptual Shortcuts: Causal-Inspired Debiasing Optimization for Generalizable Video Reasoning in Lightweight MLLMs
• Abstract summary: We propose a framework that cultivates robust reasoning in lightweight models through a two-stage debiasing process.<n>Our model, VideoThinker-R1, establishes a new state-of-the-art in video reasoning efficiency.
• Score: 11.567226738245175
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Although reinforcement learning (RL) has significantly advanced reasoning capabilities in large multimodal language models (MLLMs), its efficacy remains limited for lightweight models essential for edge deployments. To address this issue, we leverage causal analysis and experiment to reveal the underlying phenomenon of perceptual bias, demonstrating that RL-based fine-tuning compels lightweight models to preferentially adopt perceptual shortcuts induced by data biases, rather than developing genuine reasoning abilities. Motivated by this insight, we propose VideoThinker, a causal-inspired framework that cultivates robust reasoning in lightweight models through a two-stage debiasing process. First, the Bias Aware Training stage forges a dedicated "bias model" to embody these shortcut behaviors. Then, the Causal Debiasing Policy Optimization (CDPO) algorithm fine-tunes the primary model, employing an innovative repulsive objective to actively push it away from the bias model's flawed logic while simultaneously pulling it toward correct, generalizable solutions. Our model, VideoThinker-R1, establishes a new state-of-the-art in video reasoning efficiency. For same-scale comparison, requiring no Supervised Fine-Tuning (SFT) and using only 1 of the training data for RL, it surpasses VideoRFT-3B with a 3.2% average gain on widely-used benchmarks and a 7% lead on VideoMME. For cross-scale comparison, it outperforms the larger Video-UTR-7B model on multiple benchmarks, including a 2.1% gain on MVBench and a 3.8% gain on TempCompass. Code is available at https://github.com/falonss703/VideoThinker.

Related papers

• CAMEL: Confidence-Gated Reflection for Reward Modeling [26.908515245229747]
  We propose CAMEL, a confidence-gated reflection framework that performs a lightweight single-token preference decision first.<n>We train the model via reinforcement learning with counterfactual prefix augmentation, which exposes the model to diverse initial verdicts and encourages genuine revision.<n> Empirically, CAMEL achieves state-of-the-art performance on three widely used reward-model benchmarks with 82.9% average accuracy.
  arXiv  Detail & Related papers  (2026-02-24T08:20:08Z)
• Video Reasoning without Training [38.74334895026965]
  Video reasoning using Large Multimodal Models (LMMs) relies on costly reinforcement learning (RL) and verbose chain-of-thought.<n>We show that using entropy of the model's output as a signal, we discover that the high-quality models go through a series of micro-explorations and micro-exploitations.<n>We then use these novel, theoretically-grounded insights to tune the model's behavior directly at inference.
  arXiv  Detail & Related papers  (2025-10-19T23:17:13Z)
• Towards Flash Thinking via Decoupled Advantage Policy Optimization [11.025775055262569]
  Large Reasoning Models (LRMs) have achieved remarkable performance in solving complex problems via supervised fine-tuning (SFT) and reinforcement learning (RL)<n>Existing RL algorithms suffer from excessively lengthy responses and overthinking issues, resulting in increased inference latency and computational consumption.<n>We propose a novel RL framework, DEPO, to reduce inefficient reasoning for models.
  arXiv  Detail & Related papers  (2025-10-17T07:19:20Z)
• Dual-Weighted Reinforcement Learning for Generative Preference Modeling [61.443461640955796]
  We propose Dual-Weighted Reinforcement Learning (DWRL) as a new framework for preference modeling.<n>In this paper, we apply DWRL to preference modeling by training generative preference models (GPMs) to first generate a thought and then predict the human preference score.<n>Our results position DWRL as a general framework for reasoning-enhanced preference learning beyond verifiable tasks.
  arXiv  Detail & Related papers  (2025-10-17T02:14:24Z)
• VR-Thinker: Boosting Video Reward Models through Thinking-with-Image Reasoning [49.610569478718226]
  multimodal reward models (RMs) have substantially improved post-training for visual generative models.<n>VideoReward Thinker (VR-Thinker) is a thinking-with-image framework that equips the RM with visual reasoning operations and a visual memory window.<n>Our approach delivers state-of-the-art accuracy among open-source models on video preference benchmarks.
  arXiv  Detail & Related papers  (2025-10-12T09:29:50Z)
• Advancing Multimodal Reasoning via Reinforcement Learning with Cold Start [24.244577648817188]
  "aha moment" patterns are often attributed to emergent properties from reinforcement learning (RL)<n>We present a comprehensive study on enhancing multimodal reasoning through a two-stage approach.<n>Our experiments show that this combined approach consistently outperforms both SFT-only and RL-only methods.
  arXiv  Detail & Related papers  (2025-05-28T13:21:38Z)
• ViaRL: Adaptive Temporal Grounding via Visual Iterated Amplification Reinforcement Learning [68.76048244253582]
  We introduce ViaRL, the first framework to leverage rule-based reinforcement learning (RL) for optimizing frame selection in video understanding.<n>ViaRL utilizes the answer accuracy of a downstream model as a reward signal to train a frame selector through trial-and-error.<n>ViaRL consistently delivers superior temporal grounding performance and robust generalization across diverse video understanding tasks.
  arXiv  Detail & Related papers  (2025-05-21T12:29:40Z)
• Exploring the Effect of Reinforcement Learning on Video Understanding: Insights from SEED-Bench-R1 [53.894789613838654]
  We introduce SEED-Bench-R1, a benchmark designed to evaluate post-training methods for MLLMs in video understanding.<n>It includes intricate real-world videos and complex everyday planning tasks in the format of multiple-choice questions.<n>Using Qwen2-VL-Instruct-7B as a base model, we compare RL with supervised fine-tuning (SFT)<n>Our detailed analysis reveals that RL enhances visual perception but often produces less coherent reasoning chains.
  arXiv  Detail & Related papers  (2025-03-31T17:55:23Z)
• OpenVLThinker: Complex Vision-Language Reasoning via Iterative SFT-RL Cycles [91.88062410741833]
  We introduce OpenVLThinker, one of the first open-source large vision-language models (LVLMs) to exhibit sophisticated chain-of-thought reasoning.<n>We show that OpenVLThinker-7B consistently advances performance across six benchmarks demanding mathematical and general reasoning.
  arXiv  Detail & Related papers  (2025-03-21T17:52:43Z)
• R1-Zero's "Aha Moment" in Visual Reasoning on a 2B Non-SFT Model [70.77691645678804]
  We present the first successful replication of emergent characteristics for multimodal reasoning on only a non-SFT 2B model.<n>Our model achieves 59.47% accuracy on CVBench, outperforming the base model by approximately 30% and exceeding both SFT setting by 2%.<n>In addition, we share our failed attempts and insights in attempting to achieve R1-like reasoning using RL with instruct models.
  arXiv  Detail & Related papers  (2025-03-07T04:21:47Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.