Watching Itself Watch: Self-Auditing Visual Reliance for Video Reasoning
cs.CV, cs.LG
Submitted: 2026-11-09
Updated: 2026-11-09
Project page: https://vision.cs.utexas.edu/projects/VIOS
Terminology
Sources
- Qwen2.5-VL Technical Report
- Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling
- SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training
- Lost in Time: A New Temporal Benchmark for VideoLLMs
- SciVideoBench: Benchmarking Scientific Video Reasoning in Large Multimodal Models
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- Distilling the Knowledge in a Neural Network
- Video-MMMU: Evaluating Knowledge Acquisition from Multi-Discipline Professional Videos
- Scaling Laws for Neural Language Models
- Beyond Uniform Credit: Causal Credit Assignment for Policy Optimization
- Kimi k1.5: Scaling Reinforcement Learning with LLMs
- Adaptive Group Policy Optimization: Towards Stable Training and Token-Efficient Reasoning
- Video-R2: Reinforcing Consistent and Grounded Reasoning in Multimodal Language Models
- NGRPO: Negative-enhanced Group Relative Policy Optimization
- VideoMathQA: Benchmarking Mathematical Reasoning via Multimodal Understanding in Videos
- DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- Video-LMM Post-Training: A Deep Dive into Video Reasoning with Large Multimodal Models
- Exploiting Tree Structure for Credit Assignment in RL Training of LLMs
- Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs