MMMU-Pro is a stricter multimodal benchmark that removes text-only solvable questions, augments options, and requires reading text from images, yielding substantially lower model scores of 16.8-26.9%.
hub
URL https://openaccess.thecvf.com/ content_cvpr_2017/html/Goyal_Making_the_v_CVPR_2017_paper.html
13 Pith papers cite this work, alongside 2,198 external citations. Polarity classification is still indexing.
13
Pith papers citing it
2,198
external citations · OpenAlex
hub tools
citation-role summary
background 1
dataset 1
citation-polarity summary
polarities
background 2representative citing papers
MoHallBench is a new benchmark evaluating motion hallucination in VideoLLMs from co-occurrence priors, sequential inference, and similarity confusion, revealing decoupling from action recognition performance.
Brain-IT-VQA decodes visual question answers from fMRI using a transformer to extract language tokens and introduces the NSD-VQA benchmark with 20 controlled questions per image across 20 categories.
Large vision-language models exhibit severe object hallucination that varies with training instructions, and the proposed POPE polling method evaluates it more stably and flexibly than prior approaches.
Cross-scale supervision from a leakage-curated multi-magnification VQA benchmark improves pathology VLMs on both multi-image and single-image evaluation.
Deep Pre-Alignment uses a small VLM perceiver instead of ViT to pre-align visual features with LLM text space, yielding 1.9-3.0 point gains on multimodal benchmarks and 32.9% less language forgetting.
MedVIGIL provides a 300-case evaluation suite with 2556 probes that measures silent failures in medical VLMs under broken evidence, showing the best model at 69.2 on the composite score versus a human radiologist at 83.3.
SIEVES improves selective prediction coverage by up to 3x on OOD VQA benchmarks by training a selector to score the quality of visual evidence produced by reasoner models, generalizing across benchmarks and proprietary models without internal access or per-task retraining.
A query-based projector bolsters Mamba multimodal LLMs by compressing visual tokens with cross-attention without manual scan ordering.
Ask4VG learns a risk estimator from counterfactual visual probes to rerank question rewrites, reducing held-out hallucination risk from 0.658 to 0.623 and raising accuracy from 0.337 to 0.356 on VQA-RAD.
Activation verbalization methods for LLMs largely reflect the verbalizer model's parametric knowledge rather than privileged information from the target model's activations.
ABACUS adapts a 3B unified foundation model using density-aware zooming, boundary-aware GRPO, and cycle-consistent self-critique to achieve SOTA on seven counting and generation benchmarks without task-specific training.
citing papers explorer
-
MMMU-Pro: A More Robust Multi-discipline Multimodal Understanding Benchmark
MMMU-Pro is a stricter multimodal benchmark that removes text-only solvable questions, augments options, and requires reading text from images, yielding substantially lower model scores of 16.8-26.9%.
-
MoHallBench: A Benchmark for Motion Hallucination in Video Large Language Models
MoHallBench is a new benchmark evaluating motion hallucination in VideoLLMs from co-occurrence priors, sequential inference, and similarity confusion, revealing decoupling from action recognition performance.
-
Brain-IT-VQA: From Brain Signals to Answers
Brain-IT-VQA decodes visual question answers from fMRI using a transformer to extract language tokens and introduces the NSD-VQA benchmark with 20 controlled questions per image across 20 categories.
-
Evaluating Object Hallucination in Large Vision-Language Models
Large vision-language models exhibit severe object hallucination that varies with training instructions, and the proposed POPE polling method evaluates it more stably and flexibly than prior approaches.
-
Enhancing Pathological VLMs with Cross-scale Reasoning
Cross-scale supervision from a leakage-curated multi-magnification VQA benchmark improves pathology VLMs on both multi-image and single-image evaluation.
-
Deep Pre-Alignment for VLMs
Deep Pre-Alignment uses a small VLM perceiver instead of ViT to pre-align visual features with LLM text space, yielding 1.9-3.0 point gains on multimodal benchmarks and 32.9% less language forgetting.
-
MedVIGIL: Evaluating Trustworthy Medical VLMs Under Broken Visual Evidence
MedVIGIL provides a 300-case evaluation suite with 2556 probes that measures silent failures in medical VLMs under broken evidence, showing the best model at 69.2 on the composite score versus a human radiologist at 83.3.
-
SIEVES: Selective Prediction Generalizes through Visual Evidence Scoring
SIEVES improves selective prediction coverage by up to 3x on OOD VQA benchmarks by training a selector to score the quality of visual evidence produced by reasoner models, generalizing across benchmarks and proprietary models without internal access or per-task retraining.
-
Query-based Cross-Modal Projector Bolstering Mamba Multimodal LLM
A query-based projector bolsters Mamba multimodal LLMs by compressing visual tokens with cross-attention without manual scan ordering.
-
Ask4VG: Risk-Aware Question Selection for Reducing Prior-Driven Answers in Medical VQA
Ask4VG learns a risk estimator from counterfactual visual probes to rerank question rewrites, reducing held-out hallucination risk from 0.658 to 0.623 and raising accuracy from 0.337 to 0.356 on VQA-RAD.
-
Do Activation Verbalization Methods Convey Privileged Information?
Activation verbalization methods for LLMs largely reflect the verbalizer model's parametric knowledge rather than privileged information from the target model's activations.
-
ABACUS: Adapting Unified Foundation Model for Bridging Image Count Understanding and Generation
ABACUS adapts a 3B unified foundation model using density-aware zooming, boundary-aware GRPO, and cycle-consistent self-critique to achieve SOTA on seven counting and generation benchmarks without task-specific training.
- MCMit: Hardware-Software Co-Design for Mid-Circuit Measurement Error Mitigation