AMBER

The 'AMBER' dataset/benchmark is used to evaluate the reliability of vision-language models by assessing their ability to generate consistent and accurate responses to queries.

Papers using AMBER (12)

SeeMe: Mitigating Hallucinations in Large Vision-Language Models through Effective Visual Token Engineering2026When Language Overwrites Vision: Over-Alignment and Geometric Debiasing in Vision-Language Models2026Dual-Pathway Circuits of Object Hallucination in Vision-Language Models2026Instruction-Evidence Contrastive Dual-Stream Decoding for Grounded Vision-Language Reasoning2026Overthinking Causes Hallucination: Tracing Confounder Propagation in Vision Language Models2026Beyond Dominant Patches: Spatial Credit Redistribution For Grounded Vision-Language Models2026AFTER: Mitigating the Object Hallucination of LVLM via Adaptive Factual-Guided Activation Editing2026Context-Aware Decoding for Faithful Vision-Language Generation2026MaskCD: Mitigating LVLM Hallucinations by Image Head Masked Contrastive Decoding2025Watermarking for Factuality: Guiding Vision-Language Models Toward Truth via Tri-layer Contrastive Decoding2025Self-Consistency as a Free Lunch: Reducing Hallucinations in Vision-Language Models via Self-Reflection2025Mitigating Hallucination in Large Vision-Language Models via Adaptive Attention Calibration2025