Learning to Look Again: Loss-Gap Supervision for Free-form Crop Routing in Vision-Language Models

arXiv:2608.21762 · cs.CV, cs.CL · Submitted 2026-08-22 · Read on arXiv

cs.CV, cs.CL

Submitted: 2026-08-22

Updated: 2026-08-29

Terminology

• Benchmark — 30× in this paper · explained in 2 episodes
• Vision-language models — 6× in this paper · explained in 2 episodes
• Transformer — 4× in this paper · explained in 2 episodes
• Instruction Tuning — 3× in this paper · explained in 2 episodes
• Ablation — 2× in this paper · explained in 1 episode
• Core — 2× in this paper · explained in 2 episodes
• Redundancy — 2× in this paper · explained in 2 episodes
• Reinforcement Learning — 2× in this paper · explained in 4 episodes
• Vision-Language Models (VLMs) — 2× in this paper · explained in 6 episodes
• Abstention — 1× in this paper · explained in 1 episode
• Alignment — 1× in this paper · explained in 1 episode
• Consistency — 1× in this paper · explained in 3 episodes
• Cross-Model Transfer — 1× in this paper · explained in 2 episodes
• Deep Reinforcement Learning — 1× in this paper · explained in 2 episodes
• Foundation Model — 1× in this paper · explained in 3 episodes
• Foundation Models — 1× in this paper · explained in 7 episodes
• Iterative Refinement — 1× in this paper · explained in 3 episodes
• Representation Learning — 1× in this paper · explained in 2 episodes
• Test-time Scaling — 1× in this paper · explained in 2 episodes
• Trace — 1× in this paper · explained in 2 episodes

Sources

• Flamingo: a Visual Language Model for Few-Shot Learning
• Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond
• Constitutional AI: Harmlessness from AI Feedback
• Token Merging: Your ViT But Faster
• CropVLM: Learning to Zoom for Fine-Grained Vision-Language Perception
• PaLI: A Jointly-Scaled Multilingual Language-Image Model
• Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling
• Training language models to follow instructions with human feedback
• Direct Preference Optimization: Your Language Model is Secretly a Reward Model
• TokenLearner: What Can 8 Learned Tokens Do for Images and Videos?
• LLaVA-PruMerge: Adaptive Token Reduction for Efficient Large Multimodal Models
• Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution
• LLaVA-UHD: an LMM Perceiving Any Aspect Ratio and High-Resolution Images
• mPLUG-DocOwl: Modularized Multimodal Large Language Model for Document Understanding
• Zoom-Refine: Boosting High-Resolution Multimodal Understanding via Localized Zoom and Self-Refinement
• Towards Perceiving Small Visual Details in Zero-shot Visual Question Answering with Multimodal LLMs

Related papers

• Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
• AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
• Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
• MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
• TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
• A Survey on Efficient Vision-Language-Action Models