Back to the Barn with LLAMAs: Evolving Pretrained LLM Backbones in Finetuning Vision Language Models

TL;DR

This paper systematically investigates how evolving pretrained LLM backbones impact downstream vision-language model performance, revealing that newer LLMs do not always improve VLMs and their benefits vary by task.

Contribution

It provides a controlled analysis of the effects of different LLAMA LLM backbones on VLM performance, highlighting task-dependent benefits and model processing differences.

Findings

Newer LLM backbones do not always enhance VLM performance.

VLM capabilities vary with the LLM backbone, especially in reasoning tasks.

Tasks relying on visual understanding see limited gains from newer LLMs.

Abstract

Vision-Language Models (VLMs) have rapidly advanced by leveraging powerful pre-trained Large Language Models (LLMs) as core reasoning backbones. As new and more capable LLMs emerge with improved reasoning, instruction-following, and generalization, there is a pressing need to efficiently update existing VLMs to incorporate these advancements. However, the integration of new LLMs into VLMs, particularly how the evolving LLMs contribute to multimodal reasoning, alignment, and task-specific performance remains underexplored. Addressing this gap is important for VLM development, given the rapid evolution of pretrained LLM backbones. This study presents a controlled and systematic investigation of how changes in the pretrained LLM backbone affect downstream VLM task performance. By having the vision encoder, training data, and post-training algorithm remain same across LLAMA-1, LLAMA-2, and LLAMA-3 based VLMs, we find that newer LLM backbones do not always lead to better VLMs, but the performance depends on the downstream VLM task. For example, in visual question and answering tasks, newer LLM backbones tend to solve different questions rather than just more questions, and our analysis shows this is driven by differences in how the models process information, including better calibrated confidence and more stable internal representations. We also find that some VLM capabilities appear only in the newest LLM generation, while tasks that depend mainly on visual understanding see little benefit from a newer LLM backbone.