Pith. sign in

REVIEW 12 cited by

SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Model

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2406.12030 v4 pith:62PRQDHY submitted 2024-06-17 cs.CV cs.AIcs.CL

classification cs.CVcs.AIcs.CL
keywords alignmentspa-vlvlmsdatasetmodelssafetylanguagepreference
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

The emergence of Vision Language Models (VLMs) has brought unprecedented advances in understanding multimodal information. The combination of textual and visual semantics in VLMs is highly complex and diverse, making the safety alignment of these models challenging. Furthermore, due to the limited study on the safety alignment of VLMs, there is a lack of large-scale, high-quality datasets. To address these limitations, we propose a Safety Preference Alignment dataset for Vision Language Models named SPA-VL. In terms of breadth, SPA-VL covers 6 harmfulness domains, 13 categories, and 53 subcategories, and contains 100,788 samples of the quadruple (question, image, chosen response, rejected response). In terms of depth, the responses are collected from 12 open-source (e.g., QwenVL) and closed-source (e.g., Gemini) VLMs to ensure diversity. The construction of preference data is fully automated, and the experimental results indicate that models trained with alignment techniques on the SPA-VL dataset exhibit substantial improvements in harmlessness and helpfulness while maintaining core capabilities. SPA-VL, as a large-scale, high-quality, and diverse dataset, represents a significant milestone in ensuring that VLMs achieve both harmlessness and helpfulness.

Discussion (0). Sign in to comment.

Forward citations

Cited by 12 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Multimodal Reward Hacking in Reinforcement Learning

    cs.AI 2026-07 conditional novelty 6.5 of 10

    Imperfect multimodal RL rewards systematically create new failures (NRFR > RHR); scaling and answer-aware rewards help but do not eliminate hacking, and unreliable visual verifiers actively increase it.

  2. When Are Reasoning-Based Guardrails Not Efficient? ResponseGuard: A Fast Vision-Language Guard for Real-Time Moderation

    cs.CV 2026-07 conditional novelty 6.0 of 10

    A label-only 2B guard outperforms a 3B reasoning-based vision-language guard on response-harmfulness detection at about 150x lower latency, with the remaining image gap attributed to frozen encoders.

  3. GrAInS: Gradient-based Attribution for Inference-Time Steering of LLMs and VLMs

    cs.CL 2025-07 conditional novelty 6.0 of 10

    GrAInS uses Integrated Gradients to identify the most influential tokens, then builds layer-wise steering vectors that improve truthfulness, reduce hallucination, and preserve general capabilities in LLMs and VLMs.

  4. Bridging the Gap in Vision Language Models in Identifying Unsafe Concepts Across Modalities

    cs.CR 2025-07 conditional novelty 6.0 of 10

    Vision-language models consistently recognize unsafe content better from text than from images, and a simplified reinforcement learning fine-tune narrows that gap.

  5. The Man Behind the Sound: Demystifying Audio Private Attribute Profiling via Multimodal Large Language Model Agents

    cs.CR 2025-07 unverdicted novelty 6.0 of 10

    A multi-agent audio-language model framework can automatically profile private attributes, such as age, health, and income, directly from general audio recordings.

  6. GThinker: Towards General Multimodal Reasoning via Cue-Guided Rethinking

    cs.CV 2025-06 conditional novelty 6.0 of 10

    A cue-tagging and rethinking training recipe lifts a 7B multimodal model to 81.5% on M3CoT, though the gain may reflect training on the same benchmark.

  7. Bootstrapping LLM Robustness for VLM Safety via Reducing the Pretraining Modality Gap

    cs.AI 2025-05 conditional novelty 6.0 of 10

    Reducing the modality gap between image and text embeddings during LVLM pretraining reduces unsafe response rates by up to 16.3% across models and benchmarks.

  8. USB: A Comprehensive and Unified Safety Evaluation Benchmark for Multimodal Large Language Models

    cs.CR 2025-05 conditional novelty 6.0 of 10

    USB-SafeBench is a unified MLLM safety benchmark with 61 risk categories, 4 modality combinations, and dual-language vulnerability and oversensitivity tests.

  9. A Survey on Evaluating Quality and Trustworthiness in LLM-Generated Data

    cs.AI 2026-01 conditional novelty 5.0 of 10

    A metric-oriented survey that classifies intrinsic quality and trustworthiness metrics for LLM-generated data across six modalities and documents systematic evaluation gaps in the current literature.

  10. Seeing the Threat: Vulnerabilities in Vision-Language Models to Adversarial Attack

    cs.CL 2025-05 conditional novelty 5.0 of 10

    A two-stage evaluation framework and token-projection analysis show that LVLMs encode harmful semantic cues from images even without OCR, while remaining vulnerable to cross-modal attacks.

  11. VSCBench: Bridging the Gap in Vision-Language Model Safety Calibration

    cs.IR 2025-05 conditional novelty 5.0 of 10

    A new benchmark, VSCBench, measures oversafety and undersafety in vision-language models and shows that most models, including proprietary ones, are miscalibrated on at least one safety dimension.

  12. Aligning Large Vision-Language Models by Deep Reinforcement Learning and Direct Preference Optimization

    cs.LG 2025-09 unverdicted

    A survey categorizing deep reinforcement learning and direct preference optimization methods for aligning large vision-language models, with tables of studies and datasets and no new experimental result.

Pith tools