Research map / Image generation / Text-to-image diffusion
DisRM: Reward Modeling as Discriminative Prediction
Runtao Liu, Jiahao Zhan, Yuxuan GUO, Yingqing He, Chen Wei, Alan Yuille, Qifeng Chen
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 |
|---|---|---|---|
| select best of 10 samples with DisRM | SD1.5 · JourneyDB | FID: 72.06 -> 68.51 | Improved |
| fine-tune with DisRM-selected samples (SFT) | SD1.5 · JourneyDB | FID: 72.06 -> 64.98 | Improved |
| DPO with DisRM-built pairs | SD1.5 · JourneyDB | FID: 72.06 -> 63.61 | Improved |
| DPO with DisRM pairs instead of DiffusionDPO 1M labeled pairs | SD1.5 · JourneyDB | FID: 68.15 -> 63.61 | Improved |
| DisRM DPO on SDXL | SDXL · JourneyDB | FID: 62.83 -> 61.95 | Mixed |
| DisRM selection on SDXL | SDXL · JourneyDB | PickScore: 21.235 -> 21.311 | No clear effect |
| DPO with DisRM using safe PPD | SD1.5 · safety COCO | IP: 42 -> 18 | Improved |
| DPO with DisRM using safe PPD | SDXL · safety COCO | IP: 51 -> 17 | Improved |
| increase PPD size from 2K to 15K for selection (ablation) | SD1.5 · safety COCO | IP: 35 -> 34 | No clear effect |
| DPO with DisRM on VideoCrafter2 | VideoCrafter2 · VBench | VBench: 80.44 -> 81.38 | Improved |
| add rank-based bootstrapping instead of naive single checkpoint (ablation) | DisRM · selection on JourneyDB | FID: 14.48 -> 14.18 | Improved |
| annotate DPO pairs with DisRM instead of ImageReward | SD1.5 · 10k-pair DPO | FID: 72.59 -> 67.42 | 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.
- SDXL base — SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis (ICLR 2024) · better on 3 of 3 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
- RoVRM: A Robust Visual Reward Model Optimized via Auxiliary Textual Preference Data (AAAI 2025 · similarity 0.88)
- The Reward Was in Your Data All Along: Correcting Flow Matching with Discriminator-Guided RL. (NeurIPS 2026 · similarity 0.88)
- Think, then Score: Decoupled Reasoning and Scoring for Video Reward Modeling (NeurIPS 2026 · similarity 0.87)
- Auto-Rubric as Reward: From Implicit Preference to Explicit Generative Criteria (NeurIPS 2026 · similarity 0.87)
- Reward-Instruct: A Reward-Centric Approach to Fast Photo-Realistic Image Generation (NeurIPS 2025 · similarity 0.87)
- DrPO: Drifting Preference Optimization for One-Step Generative Models (NeurIPS 2026 · similarity 0.86)
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.
