# Synth-Align: Improving Trustworthiness in Vision-Language Model with Synthetic Preference Data Alignment

arXiv: https://arxiv.org/abs/2412.17417 · HTML: https://mlsanity.com/paper/2412.17417/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 31 of its 43 references that are in the map.
- First posted on arXiv 23 Dec 2024.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM training and visual instruction tuning](https://mlsanity.com/mllm/mllm-training-and-visual-instruction-tuning/)
- LLM alignment / RLHF / preference optimisation / [Diffusion and multimodal model alignment](https://mlsanity.com/alignment/diffusion-and-multimodal-model-alignment/)

## Builds on (most cited first, 30 of 31)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://mlsanity.com/paper/2305.06500/) (2023)
- [InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks](https://mlsanity.com/paper/2312.14238/) (2023)
- [Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond](https://mlsanity.com/paper/2308.12966/) (2023)
- [MMMU: A Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark for Expert AGI](https://mlsanity.com/paper/2311.16502/) (2023)
- [Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering](https://mlsanity.com/paper/2209.09513/) (2022)
- [MM-Vet: Evaluating Large Multimodal Models for Integrated Capabilities](https://mlsanity.com/paper/2308.02490/) (2023)
- [LIMA: Less Is More for Alignment](https://mlsanity.com/paper/2305.11206/) (2023)
- [ShareGPT4V: Improving Large Multi-Modal Models with Better Captions](https://mlsanity.com/paper/2311.12793/) (2023)
- [SEED-Bench: Benchmarking Multimodal LLMs with Generative Comprehension](https://mlsanity.com/paper/2307.16125/) (2023)
- [Shikra: Unleashing Multimodal LLM's Referential Dialogue Magic](https://mlsanity.com/paper/2306.15195/) (2023)
- [VizWiz Grand Challenge: Answering Visual Questions from Blind People](https://mlsanity.com/paper/1802.08218/) (2018)
- [Aligning Large Multimodal Models with Factually Augmented RLHF](https://mlsanity.com/paper/2309.14525/) (2023)
- [Object Hallucination in Image Captioning](https://mlsanity.com/paper/1809.02156/) (2018)
- [RLHF-V: Towards Trustworthy MLLMs via Behavior Alignment from Fine-grained Correctional Human Feedback](https://mlsanity.com/paper/2312.00849/) (2023)
- [Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts](https://mlsanity.com/paper/2406.12845/) (2024)
- [What Makes Good Data for Alignment? A Comprehensive Study of Automatic Data Selection in Instruction Tuning](https://mlsanity.com/paper/2312.15685/) (2023)
- [The Curse of Recursion: Training on Generated Data Makes Models Forget](https://mlsanity.com/paper/2305.17493/) (2023)
- [AMBER: An LLM-free Multi-dimensional Benchmark for MLLMs Hallucination Evaluation](https://mlsanity.com/paper/2311.07397/) (2023)
- [MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training](https://mlsanity.com/paper/2403.09611/) (2024)
- [Aligning Modalities in Vision Large Language Models via Preference Fine-tuning](https://mlsanity.com/paper/2402.11411/) (2024)
- [SVIT: Scaling up Visual Instruction Tuning](https://mlsanity.com/paper/2307.04087/) (2023)
- [Efficient Multimodal Learning from Data-centric Perspective](https://mlsanity.com/paper/2402.11530/) (2024)
- [CapsFusion: Rethinking Image-Text Data at Scale](https://mlsanity.com/paper/2310.20550/) (2023)
- [VeCLIP: Improving CLIP Training via Visual-enriched Captions](https://mlsanity.com/paper/2310.07699/) (2023)
- [Will Large-scale Generative Models Corrupt Future Datasets?](https://mlsanity.com/paper/2211.08095/) (2022)

## Built on by (most cited first, 6 of 6)

- [A Comprehensive Survey of Reward Models: Taxonomy, Applications, Challenges, and Future](https://mlsanity.com/paper/2504.12328/) (2025)
- [Aligning Multimodal LLM with Human Preference: A Survey](https://mlsanity.com/paper/2503.14504/) (2025)
- [ARM-Thinker: Reinforcing Multimodal Generative Reward Models with Agentic Tool Use and Visual Reasoning](https://mlsanity.com/paper/2512.05111/) (2025)
- [TableDreamer: Progressive and Weakness-guided Data Synthesis from Scratch for Table Instruction Tuning](https://mlsanity.com/paper/2506.08646/) (2025)
- [Transferring Textual Preferences to Vision-Language Understanding through Model Merging](https://mlsanity.com/paper/2502.13487/) (2025)
- [NodeSynth: Socially Aligned Synthetic Data for AI Evaluation](https://mlsanity.com/paper/2605.14381/) (2026)

## Abstract (arXiv)

Large Vision-Language Models (LVLMs) have shown promising capabilities in understanding and generating information by integrating both visual and textual data. However, current models are still prone to hallucinations, which degrade the performance and greatly harm the user experience in real-world applications. Post-training alignment, particularly preference-tuning, is intended to align model outputs and behaviors (safety, instruction-following, style), ensuring robustness and adaptability to a wide range of tasks. The use of synthetic data for alignment, particularly in multimodal settings, remains under explored. Existing approaches typically use a strong model or a ground-truth model (CLIP) to determine positive and negative image-text data points. This paper proposes SynthAlign, a pipeline to generate and collect synthetic human-preference image-text data with optimal control built specifically for post-training alignment with DPO. At the core of the framework is the utilization of reward models as a proxy of human preference. A series of evaluation and benchmarking is provided to validate the effectiveness of the proposed framework and the resulting dataset. Notably, our framework enhanced LLaVA-1.5-7B achieved substantial POPE improvements: 87.6% accuracy and 97.8% precision, MMHal-Bench score increased from 2.36 to 3.49, and hallucination rate decreased from 51.0% to 25.0% (a 50.98% relative reduction).