image tokens

4 papers

• Grounding Language with Vision: A Conditional Mutual Information Calibrated Decoding Strategy for Reducing Hallucinations in LVLMs
• Polyline Path Masked Attention for Vision Transformer
• The Narrow Gate: Localized Image-Text Communication in Native Multimodal Models
• ViSpec: Accelerating Vision-Language Models with Vision-Aware Speculative Decoding