Research map / Image generation / Text-to-image diffusion
TEMPFLOW-GRPO: WHEN TIMING MATTERS FOR GRPO IN FLOW MODELS
Xiaoxuan He, Siming Fu, Yuke Zhao, Wanli Li, Jian Yang, Dacheng Yin, Fengyun Rao, Bo Zhang
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 |
|---|---|---|---|
| replace Flow-GRPO with TempFlow-GRPO | SD3.5-M · GenEval | overall score: 0.88 -> 0.97 at 3800 steps | Improved |
| TempFlow-GRPO convergence speed | SD3.5-M · GenEval | steps to 0.95: about 2,000 vs 5,600 steps | Improved |
| replace Flow-GRPO with TempFlow-GRPO | SD3.5-M · PickScore | PickScore: +1.7% vs Flow-GRPO, +1.0% vs Flow-GRPO (Prompt) | Improved |
| replace Flow-GRPO with TempFlow-GRPO | FLUX.1-dev · HPSv3 reward | steps to match: 300 -> 80 steps | Improved |
| add noise-aware policy reweighting (ablation) | Flow-GRPO · GenEval | score at 1200 steps: 0.82 -> 0.92 | Improved |
| add trajectory branching (ablation) | Flow-GRPO · GenEval | score: approx +5% | Improved |
| add seed group strategy (ablation) | TempFlow-GRPO · GenEval | score: approx +2% | Improved |
| add trajectory branching, noise reweighting, then seed group (ablation) | Flow-GRPO (Prompt) · PickScore | PickScore | Improved |
| more initial noise seeds with fewer branches (6x4 vs 2x12) (ablation) | TempFlow-GRPO · group size 24 | training curve | Mixed |
| replace Flow-GRPO (Prompt) with TempFlow-GRPO under multi-reward | FLUX.1-dev · multi-reward PickScore prompts | HPSv3: 10.41 -> 12.23 | Improved |
| replace DanceGRPO with TempFlow-GRPO | HPSv2 reward · HPDv2 | HPSv2 score: 37.2 -> 38.5 | Improved |
| use TempFlow-GRPO at 1024 resolution | SD3.5-M 1024 · PickScore | steps to match: 450 -> about 100 steps | Improved |
Later papers that compared with this method
1 later accepted paper reports TempFlow-GRPO as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Alleviating Sparse Rewards by Modeling Step-Wise and Long-Term Sampling Effects in Flow-Based GRPO (ICML 2026 · better on 9 of 9 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
- Flow-GRPO: Training Flow Matching Models via Online RL (NeurIPS 2025 · similarity 0.93)
- From Sparse to Dense: Multi-View GRPO for Flow Models via Augmented Condition Space (ECCV 2026 · similarity 0.92)
- MixGRPO: Unlocking Flow-based GRPO Efficiency with Mixed ODE-SDE (ECCV 2026 · similarity 0.91)
- Test-Time Reinforcement Learning for Flow Matching (ICML 2026 · similarity 0.91)
- Reinforcement Learning Meets Masked Generative Models: Mask-GRPO for Text-to-Image Generation (NeurIPS 2025 · similarity 0.90)
- TMPO: Trajectory Matching Policy Optimization for Diverse and Efficient Diffusion Alignment (NeurIPS 2026 · similarity 0.89)
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.
