Paper Detail
Yang Yang, Qinyu Zhao, Mouxiang Chen, Xiaohui Li, Lixin Gu, Wenhai Wang, Hongjie Zhang, Wenwei Zhang
Existing scaling strategies for Multimodal Large Language Models (MLLMs) typically expand either model parameters or sequential inference computation, incurring substantial memory or latency overhead. More importantly, most existing methods fail to alter the rigid, fixed computation allocation between the Vision Transformer and the Large Language Model components, limiting task-specific optimization. To address this, we introduce the Parallel Vision-Language (ParVL) scaling framework for MLLMs, which scales parallel computation by reusing the existing ViT and LLM backbone parameters across multiple vision and language branches. This framework raises a central question: given a fixed backbone parameter budget, how should additional shared-backbone computation be allocated between the vision and language modalities? We instantiate each parallel computational stream with branch-specific prefix parameters over a shared backbone, and train the entire model end-to-end via full-parameter supervised fine-tuning on roughly 13B tokens. We systematically study the computation-allocation trade-off between the ViT encoder and LLM decoder. ParVL improves overall multimodal performance over same-recipe single-branch baselines, and the best evaluated vision--language allocation varies across tasks. Code is available at https://github.com/YangYangGirl/ParVL.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@article{yang2026parvl,
title = {ParVL: Parallel Scaling and Expandable Compute Allocation for Multimodal LLMs},
author = {Yang Yang and Qinyu Zhao and Mouxiang Chen and Xiaohui Li and Lixin Gu and Wenhai Wang and Hongjie Zhang and Wenwei Zhang},
year = {2026},
abstract = {Existing scaling strategies for Multimodal Large Language Models (MLLMs) typically expand either model parameters or sequential inference computation, incurring substantial memory or latency overhead. More importantly, most existing methods fail to alter the rigid, fixed computation allocation between the Vision Transformer and the Large Language Model components, limiting task-specific optimization. To address this, we introduce the Parallel Vision-Language (ParVL) scaling framework for MLLMs, },
url = {https://arxiv.org/abs/2608.04010},
keywords = {cs.CV, cs.CL},
eprint = {2608.04010},
archiveprefix = {arXiv},
}
{}