Research map / Vision-language and multimodal / Multimodal LLMs
Scaling Capability in Token Space: An Analysis of Large Vision Language Model
Tenghui Li, Guoxu Zhou, Xuyang Zhao, Qibin Zhao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase the number of learnable queries n_l (vision tokens) (ablation) | Vision Question Queries model · OCRBench | fitted scaling exponent alpha: alpha -0.3727 | Improved |
| increase n_l (ablation) | Vision Queries (ft) model · COCO VAL CIDEr | fitted scaling exponent alpha: alpha -0.1642 -> -0.3937 | Improved |
| increase n_l (ablation) | Vision Question Queries model · RealWorldQA | fitted scaling exponent alpha: alpha -0.0030 | No clear effect |
| increase n_l (ablation) | Vision Queries (ft) model · RealWorldQA | fitted scaling exponent alpha: alpha 0.0335 | No clear effect |
| increase n_l (ablation) | Vision Question Queries model · MMStar | fitted scaling exponent alpha: alpha -0.0272 | Improved |
| increase n_l (ablation) | Vision Question Queries model · MME | fitted scaling exponent alpha: alpha -0.0516 | Improved |
| increase n_l (ablation) | Vision Question Queries model · COCO VAL BLEU-4 | fitted scaling exponent alpha: alpha -0.1161 | Improved |
| remove question text from fusion input (ft vision queries only) (ablation) | Vision Queries (ft) model · OCRBench | fitted scaling exponent alpha: alpha -0.3727 -> -0.4142 | Improved |
| remove question text from fusion input (ft vision queries only) (ablation) | Vision Queries (ft) model · POPE | fitted scaling exponent alpha: alpha -0.0503 -> -0.0377 | Hurt |
| remove question text from fusion input (ft vision queries only) (ablation) | Vision Queries (ft) model · AI2D TEST | fitted scaling exponent alpha: alpha -0.0359 -> -0.0902 | Improved |
| remove question text from fusion input (ft vision queries only) (ablation) | Vision Queries (ft) model · SEEDBench IMG | fitted scaling exponent alpha: alpha -0.0710 -> -0.0516 | Hurt |
Closest papers by meaning
- Benchmarking Large Vision-Language Models on Fine-Grained Image Tasks: A Comprehensive Evaluation (ICLR 2026 · similarity 0.89)
- Synthesize Diagnose and Optimize: Towards Fine-Grained Vision-Language Understanding (CVPR 2024 · similarity 0.89)
- Scaling Laws for Robust Comparison of Open Foundation Language-Vision Models and Datasets (NeurIPS 2025 · similarity 0.89)
- Towards Interpreting Visual Information Processing in Vision-Language Models (ICLR 2025 · similarity 0.89)
- SCoRe: Salience-Coverage Reduction for Vision Token Pruning in Vision-Language Models (CVPR 2026 · similarity 0.89)
- What matters when building vision-language models? (NeurIPS 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
