Prioritizing Image-Related Tokens Enhances Vision-Language Pre-Training

TL;DR

PRIOR is a novel vision-language pre-training method that improves model accuracy by emphasizing image-related tokens during training, reducing noise and hallucination, and demonstrating significant performance gains.

Contribution

It introduces a token re-weighting approach based on importance sampling using a text-only reference model, enhancing vision-language pre-training effectiveness.

Findings

19% average relative improvement on benchmarks

8% average relative improvement without visual encoders

Higher scaling coefficients indicating better potential for future gains

Abstract

In standard large vision-language models (LVLMs) pre-training, the model typically maximizes the joint probability of the caption conditioned on the image via next-token prediction (NTP); however, since only a small subset of caption tokens directly relates to the visual content, this naive NTP unintentionally fits the model to noise and increases the risk of hallucination. We present PRIOR, a simple vision-language pre-training approach that addresses this issue by prioritizing image-related tokens through differential weighting in the NTP loss, drawing from the importance sampling framework. PRIOR introduces a reference model-a text-only large language model (LLM) trained on the captions without image inputs, to weight each token based on its probability for LVLMs training. Intuitively, tokens that are directly related to the visual inputs are harder to predict without the image and thus receive lower probabilities from the text-only reference LLM. During training, we implement a token-specific re-weighting term based on the importance scores to adjust each token's loss. We implement PRIOR in two distinct settings: LVLMs with visual encoders and LVLMs without visual encoders. We observe 19% and 8% average relative improvement, respectively, on several vision-language benchmarks compared to NTP. In addition, PRIOR exhibits superior scaling properties, as demonstrated by significantly higher scaling coefficients, indicating greater potential for performance gains compared to NTP given increasing compute and data.