Research map / Image generation / Text-to-image diffusion
Personalized Reward Modeling for Text-to-Image Generation
Jeongeun Lee, Ryang Heo, Dongha Lee
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 |
|---|---|---|---|
| add CoT reasoning (ablation) | PIGReward · PIGBench | accuracy: 39.62 -> 70.83 | Improved |
| add CoT reasoning (ablation) | PIGReward · Pick-a-Pic | accuracy: 44.43 -> 45.63 | Improved |
| add DPO tuning of preference reasoner (ablation) | PIGReward · Pick-a-Pic | accuracy: 49.98 -> 63.76 | Improved |
| add DPO tuning of preference reasoner (ablation) | PIGReward · PIGBench | accuracy: 58.49 -> 84.91 | Improved |
| add CoT-distilled SFT reward model (ablation) | PIGReward · Pick-a-Pic | accuracy: 45.63 -> 63.76 | Improved |
| add CoT-distilled SFT reward model (ablation) | PIGReward · PIGBench | accuracy: 70.83 -> 84.91 | Improved |
| add SFT reward model without DPO (ablation) | PIGReward · Pick-a-Pic | accuracy: 48.94 -> 49.98 | No clear effect |
| increase user reference size from 1 to 8 (ablation) | PIGReward · Pick-a-Pic | accuracy | Improved |
| use PIGReward with user references instead of GPT-4o with references | PIGReward · PIGBench | acc w/ tie: 64.15 -> 84.91 | Improved |
| use personalized prompts optimized by PIGReward instead of Promptist | SDXL · PIGBench prompt optimization | win rate | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ImageReward — ImageReward: Learning and Evaluating Human Preferences for Text-to-Image Generation (NeurIPS 2023) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Tailored Visions: Enhancing Text-to-Image Generation with Personalized Prompt Rewriting (CVPR 2024 · similarity 0.88)
- Yo'Chameleon: Personalized Vision and Language Generation (CVPR 2025 · similarity 0.88)
- Multimodal Large Language Models Make Text-to-Image Generative Models Align Better (NeurIPS 2024 · similarity 0.88)
- MM-R1: Unleashing the Power of Unified Multimodal Large Language Models for Personalized Image Generation (AAAI 2026 · similarity 0.88)
- Preference Adaptive and Sequential Text-to-Image Generation (ICML 2025 · similarity 0.87)
- Subject-driven Text-to-Image Generation via Preference-based Reinforcement Learning (NeurIPS 2024 · similarity 0.87)
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.
