Fine-tuned Qwen2-VL-2B-Instruct model for image aesthetic assessment using the RvTC (Regression via Transformer-Based Classification) framework. This checkpoint uses image-only training without textual context.

Base Model

Architecture: Qwen2-VL-2B-Instruct
Source: Qwen/Qwen2-VL-2B-Instruct

Training Configuration

Dataset: AVA (Aesthetic Visual Analysis)
Training Mode: Image-only (no textual prompts)
Epochs: 2
Learning Rate: 1e-5
Batch Size: 128 (training)
Optimizer: AdamW with cosine scheduler
Warmup Ratio: 0.03

Binning Configuration

Number of Bins: 51
Value Range: [1.81, 8.60] (range of train set)
Method: Uniform binning for regression via classification

Performance

Evaluated on AVA test set (19,930 samples):

Pearson Correlation (PLCC): 0.841
Spearman Correlation (SRCC): 0.843

Citation

bibtex

@inproceedings{jennings2025language,
  title={Language Integration in Fine-Tuning Multimodal Large Language Models for Image-Based Regression},
  author={Roy H. Jennings, Genady Paikin, Roy Shaul, and Evgeny Soloveichik},
  booktitle={2026 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
  year={2026},
  organization={IEEE}
}

royhj/rvtc-qwen2vl-2b-image-only

Regression via Transformer-Based Classification (RvTC)

Model Description

Base Model

Training Configuration

Binning Configuration

Performance

Citation

royhj/
rvtc-qwen2vl-2b-image-only