Pith. sign in

REVIEW 6 cited by

Silkie: Preference Distillation for Large Visual Language Models

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

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 2312.10665 v1 pith:KQA7DYW6 submitted 2023-12-17 cs.CV cs.CL

Lei Li , Zhihui Xie , Mukai Li , Shunian Chen , Peiyi Wang , Liang Chen , Yazheng Yang , Benyou Wang
show 1 more author
Lingpeng Kong
This is my paper
classification cs.CVcs.CL
keywords preferencelvlmsmodelssilkievisualbenchmarkcognitiondataset
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

This paper explores preference distillation for large vision language models (LVLMs), improving their ability to generate helpful and faithful responses anchoring the visual context. We first build a vision-language feedback (VLFeedback) dataset utilizing AI annotation. Specifically, responses are generated by models sampled from 12 LVLMs, conditioned on multi-modal instructions sourced from various datasets. We adopt GPT-4V to assess the generated outputs regarding helpfulness, visual faithfulness, and ethical considerations. Furthermore, the preference supervision is distilled into Qwen-VL-Chat through the direct preference optimization (DPO) method. The resulting model Silkie, achieves 6.9% and 9.5% relative improvement on the MME benchmark regarding the perception and cognition capabilities, respectively. Silkie also demonstrates reduced hallucination by setting a new state-of-the-art score of 3.02 on the MMHal-Bench benchmark. Further analysis shows that DPO with our VLFeedback dataset mainly boosts the fine-grained perception and complex cognition abilities of LVLMs, leading to more comprehensive improvements compared to human-annotated preference datasets.

Discussion (0). Sign in to comment.

Forward citations

Cited by 6 Pith papers

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

  1. Improving Large Vision and Language Models by Learning from a Panel of Peers

    cs.CV 2025-09 conditional novelty 6.0 of 10

    A panel of LVLMs that generate, evaluate, and learn from each other's outputs improves average benchmark scores by 9 points across 15 tasks.

  2. 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.

  3. Magic-MM-Embedding: Towards Visual-Token-Efficient Universal Multimodal Embedding with MLLMs

    cs.CV 2026-02 conditional novelty 5.0 of 10

    Visual token compression (4x fewer tokens) plus a three-stage generative/contrastive/judge-curated training pipeline yields state-of-the-art MLLM-based retrieval accuracy at lower inference cost.

  4. PostAlign: Multimodal Grounding as a Corrective Lens for MLLMs

    cs.CV 2025-06 conditional novelty 5.0 of 10

    MMGrounded-PostAlign trains MLLMs to produce a grounded object token or a rejection token plus selective rationales, improving hallucination and VQA benchmarks.

  5. LeanPO: Lean Preference Optimization for Likelihood Alignment in Video-LLMs

    cs.CV 2025-06 conditional novelty 4.0 of 10

    LeanPO improves Video-LLM alignment by using a reference-free average-likelihood reward, self-generated winning/losing pairs, and dynamic label smoothing, yielding gains on six video benchmarks.

  6. Empowering Multimodal LLMs with External Tools: A Comprehensive Survey

    cs.CV 2025-08 unverdicted novelty 2.0 of 10

    A survey paper maps how external tools are used to augment multimodal large language models across data, tasks, evaluation, and future directions.

Pith tools