VLRewardBench: A Challenging Benchmark for Vision-Language Generative Reward Models
- URL: http://arxiv.org/abs/2411.17451v1
- Date: Tue, 26 Nov 2024 14:08:34 GMT
- Title: VLRewardBench: A Challenging Benchmark for Vision-Language Generative Reward Models
- Authors: Lei Li, Yuancheng Wei, Zhihui Xie, Xuqing Yang, Yifan Song, Peiyi Wang, Chenxin An, Tianyu Liu, Sujian Li, Bill Yuchen Lin, Lingpeng Kong, Qi Liu,
- Abstract summary: Vision-language generative reward models (VL-GenRMs) play a crucial role in aligning and evaluating multimodal AI systems.
Current assessment methods rely on AI-annotated preference labels from traditional tasks.
We introduce VL-RewardBench, a benchmark spanning general multimodal queries, visual hallucination detection, and complex reasoning tasks.
- Score: 66.56298924208319
- License:
- Abstract: Vision-language generative reward models (VL-GenRMs) play a crucial role in aligning and evaluating multimodal AI systems, yet their own evaluation remains under-explored. Current assessment methods primarily rely on AI-annotated preference labels from traditional VL tasks, which can introduce biases and often fail to effectively challenge state-of-the-art models. To address these limitations, we introduce VL-RewardBench, a comprehensive benchmark spanning general multimodal queries, visual hallucination detection, and complex reasoning tasks. Through our AI-assisted annotation pipeline combining sample selection with human verification, we curate 1,250 high-quality examples specifically designed to probe model limitations. Comprehensive evaluation across 16 leading large vision-language models, demonstrates VL-RewardBench's effectiveness as a challenging testbed, where even GPT-4o achieves only 65.4% accuracy, and state-of-the-art open-source models such as Qwen2-VL-72B, struggle to surpass random-guessing. Importantly, performance on VL-RewardBench strongly correlates (Pearson's r > 0.9) with MMMU-Pro accuracy using Best-of-N sampling with VL-GenRMs. Analysis experiments uncover three critical insights for improving VL-GenRMs: (i) models predominantly fail at basic visual perception tasks rather than reasoning tasks; (ii) inference-time scaling benefits vary dramatically by model capacity; and (iii) training VL-GenRMs to learn to judge substantially boosts judgment capability (+14.7% accuracy for a 7B VL-GenRM). We believe VL-RewardBench along with the experimental insights will become a valuable resource for advancing VL-GenRMs.
Related papers
- Multimodal RewardBench: Holistic Evaluation of Reward Models for Vision Language Models [82.92771279118888]
We introduce Multimodal RewardBench, an expert-annotated benchmark for evaluating multimodal reward models.
Our dataset comprises 5,211 annotated (prompt, chosen response, rejected response) triplets collected from various vision-language models.
We find that even the top-performing models, Gemini 1.5 Pro and Claude 3.5 Sonnet, achieve only 72% overall accuracy.
arXiv Detail & Related papers (2025-02-20T01:48:13Z) - AVTrustBench: Assessing and Enhancing Reliability and Robustness in Audio-Visual LLMs [70.4578433679737]
We introduce Audio-Visual Trustworthiness assessment Benchmark (AVTrustBench), comprising 600K samples spanning over 9 meticulously crafted tasks.
Using our benchmark we extensively evaluate 13 state-of-the-art AVLLMs.
The findings reveal that the majority of existing models fall significantly short of achieving human-like comprehension.
arXiv Detail & Related papers (2025-01-03T23:03:24Z) - TraceVLA: Visual Trace Prompting Enhances Spatial-Temporal Awareness for Generalist Robotic Policies [95.30717188630432]
We introduce visual trace prompting to facilitate VLA models' spatial-temporal awareness for action prediction.
We develop a new TraceVLA model by finetuning OpenVLA on our own collected dataset of 150K robot manipulation trajectories.
We present a compact VLA model based on 4B Phi-3-Vision, pretrained on the Open-X-Embodiment and finetuned on our dataset.
arXiv Detail & Related papers (2024-12-13T18:40:51Z) - Vision Language Models are In-Context Value Learners [89.29486557646624]
We present Generative Value Learning (GVL), a universal value function estimator that leverages the world knowledge embedded in vision-language models (VLMs) to predict task progress.
Without any robot or task specific training, GVL can in-context zero-shot and few-shot predict effective values for more than 300 distinct real-world tasks.
arXiv Detail & Related papers (2024-11-07T09:17:50Z) - What Are We Measuring When We Evaluate Large Vision-Language Models? An Analysis of Latent Factors and Biases [87.65903426052155]
We perform a large-scale transfer learning experiment aimed at discovering latent vision-language skills from data.
We show that generation tasks suffer from a length bias, suggesting benchmarks should balance tasks with varying output lengths.
We present a new dataset, OLIVE, which simulates user instructions in the wild and presents challenges dissimilar to all datasets we tested.
arXiv Detail & Related papers (2024-04-03T02:40:35Z) - VLUE: A Multi-Task Benchmark for Evaluating Vision-Language Models [21.549122658275383]
Recent advances in vision-language pre-training have demonstrated impressive performance in a range of vision-language tasks.
We introduce the Vision-Language Understanding Evaluation benchmark, a multi-task multi-dimension benchmark for evaluating the generalization capabilities and the efficiency-performance trade-off.
arXiv Detail & Related papers (2022-05-30T16:52:30Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.