We lifted 9 functions out of this paper's own repositories and ran 4 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| yifan123/flow_grpo | canonical | 4 of 9 |
| Function | Status | Where it lives |
|---|---|---|
| get_image_transform | Ran | yifan123/flow_grpo/flow_grpo/clip_scorer.py code served (permissive licence) · get_code("1d0c29e5ba66b093") |
| get_size | Ran | yifan123/flow_grpo/flow_grpo/clip_scorer.py code served (permissive licence) · get_code("b4ae73baea6f9f47") |
| pil_image_to_base64 | Ran | yifan123/flow_grpo/flow_grpo/qwenvl.py code served (permissive licence) · get_code("679c61d741aabb1d") |
| register_optimizer_offload_hooks | Ran | yifan123/flow_grpo/flow_grpo/fsdp_utils.py code served (permissive licence) · get_code("9f8c8299498f223d") |
| clip_score | Not yet run | yifan123/flow_grpo/flow_grpo/rewards.py code served (permissive licence) · get_code("58a9aa750111b5f3") |
| extract_scores | Not yet run | yifan123/flow_grpo/flow_grpo/qwenvl.py code served (permissive licence) · get_code("aaf3538ea48a524d") |
| fsdp_wrapper | Not yet run | yifan123/flow_grpo/flow_grpo/fsdp_utils.py code served (permissive licence) · get_code("597a05808a506d45") |
| image_similarity_score | Not yet run | yifan123/flow_grpo/flow_grpo/rewards.py code served (permissive licence) · get_code("6bfe651b4ae64450") |
| pickscore_score | Not yet run | yifan123/flow_grpo/flow_grpo/rewards.py code served (permissive licence) · get_code("667f709586d6128e") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
We propose Flow-GRPO, the first method to integrate online policy gradient reinforcement learning (RL) into flow matching models. Our approach uses two key strategies: (1) an ODE-to-SDE conversion that transforms a deterministic Ordinary Differential Equation (ODE) into an equivalent Stochastic Differential Equation (SDE) that matches the original model's marginal distribution at all timesteps, enabling statistical sampling for RL exploration; and (2) a Denoising Reduction strategy that reduces training denoising steps while retaining the original number of inference steps, significantly improving sampling efficiency without sacrificing performance. Empirically, Flow-GRPO is effective across multiple text-to-image tasks. For compositional generation, RL-tuned SD3.5-M generates nearly perfect object counts, spatial relations, and fine-grained attributes, increasing GenEval accuracy from $63\%$ to $95\%$. In visual text rendering, accuracy improves from $59\%$ to $92\%$, greatly enhancing text generation. Flow-GRPO also achieves substantial gains in human preference alignment. Notably, very little reward hacking occurred, meaning rewards did not increase at the cost of appreciable image quality or diversity degradation.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2505.05470")
get_code_for_paper("2505.05470")
have("2505.05470")
Connect an agent — have() is free.