# VeRL-omni adds day-0 RL training support for MiniMax-H3-Image

> Original title: Day-0 RL support for MiniMax-H3-Image in VeRL-omni. We train a LoRA with DiffusionNFT, using FSDP2 for training and vLLM-Omni for rollout...

- Source: vLLM
- Published: 2026-11-10T15:43:14.000Z
- Bad Influence: https://badinfluence.news/items/w76o98nhreuilaqstovsqmxy3
- Original: https://x.com/vllm_project/status/2120180487931261321

## Summary

VeRL-omni adds day-0 reinforcement learning support for MiniMax-H3-Image, training a LoRA with DiffusionNFT on the model's native 6-step distilled sampler. FSDP2 handles training and vLLM-Omni handles rollout. The reward multiplies an OCR score, a layout guard, and an HPSv2.1 quality gate, and the recipe is in pull request #890 on verl-project/verl-omni.

## Post

Day-0 RL support for MiniMax-H3-Image in VeRL-omni. We train a LoRA with DiffusionNFT, using FSDP2 for training and vLLM-Omni for rollout, directly on the model’s native 6-step distilled sampler.  The reward multiplies three terms: OCR score, a layout guard, and an HPSv2.1 quality gate. The pr is in [recipe, diffusion, fsdp] feat: add MiniMax H3 image DiffusionNFT OCR LoRA recipe by aoshen02 · Pull Request #890 · verl-project/verl-omni.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWxk6vrbsAABEW7.jpg&mode=thumb&exp=1794528000&sig=8382775b931cfe20)