VisualQuality R1 7B Our Paper has been accept as spotlight in NeurIPS 2025! This is the latest version of VisualQuality R1, trained on a diverse combination of synthetic and realistic datasets. Paper link: arXiv Code link: github The first NR IQA model enhanced by RL2R, capable of both quality description and rating through reasoning. ⚡Quick Start Non Thinking Inference When you execute inference with VisualQuality R1 as a reward/evaluation model, you can only use non thinking mode to reduce inference time, generating only a single output token with the following prompt: For single image quality rating, the code is: Example Code (VisualQuality R1: Image Quality Rating with non thinking mode) Example Code (VisualQuality R1: Batch Images Quality Rating with non thinking mode) Thinking mode for inference Example Code (VisualQuality R1: Single Image Quality Rating with thinking) Example Code (VisualQuality R1: Batch Images Quality Rating with thinking) 🚀 Updated: VisualQuality R1 high efficiency inference script with vLLM Example Code (VisualQuality R1: Batch Images Quality Rating with thinking, using vLLM) Training Preparation 1. To smoothly execute the training procedure, first down…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy