CodeGoat24/UnifiedReward-Edit-qwen35-9b

VISIONPricing:Input $0.431 / Cached $0.0862 / Output $1.12Concurrent Unit Cost:1Model Size:9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Mar 7, 2026License:mitArchitecture:Transformer Open Weights Featherless Exclusive Cold

CodeGoat24/UnifiedReward-Edit-qwen35-9b is a unified reward model developed by CodeGoat24, specifically designed for evaluating both Text-to-Image and Image-to-Image generation tasks. This model excels at assessing image editing quality through various methods including pairwise ranking, pairwise scoring, and pointwise scoring based on instruction-following and overall image quality. It is optimized for providing comprehensive feedback on generated and edited images, leveraging training data from EditScore and EditReward datasets.

Loading preview...

UnifiedReward-Edit-qwen35-9b Overview

CodeGoat24/UnifiedReward-Edit-qwen35-9b is a specialized reward model developed by CodeGoat24, focusing on the evaluation of both text-to-image and image-to-image generation. Unlike general-purpose language models, its core function is to provide nuanced feedback on the quality and adherence of generated or edited images to given instructions.

Key Capabilities

  • Unified Reward System: Evaluates both Text-to-Image and Image-to-Image generation tasks.
  • Image Editing Assessment: Supports multiple evaluation methodologies for image editing:
    • Pairwise Rank: Determines which of two edited images is superior.
    • Pairwise Score: Assigns individual scores to each image within a pair.
    • Pointwise Score: Rates a single image on two distinct axes: instruction-following and overall image quality.
  • Data-Driven Training: Utilizes preprocessed data from established datasets like EditScore and EditReward to enhance its evaluative capabilities.

Good For

  • Automated Image Quality Evaluation: Ideal for developers and researchers needing to automatically assess the quality of images generated or edited by other AI models.
  • Fine-tuning Image Generation Models: Can be used as a reward signal to improve the performance of text-to-image and image-to-image synthesis models.
  • Research in Generative AI: Provides a robust tool for analyzing and comparing different image generation and editing techniques based on objective metrics.

Finetunes
1 models