Could you provide the model trained by GRPO?

[Ethereal-sakura]

First of all, fantastic work! Thank you for your valuable contribution to the open source community.

I'm interested in the GRPO training process but currently lack sufficient GPU resources to conduct the training myself. In this regard, I have a few questions:

1. Are there any GRPO-trained models available for public use，I didn't find it on huggingface。

2. I'm particularly curious about the model's inference process after GRPO training, specifically:

   • Whether reflection capabilities can emerge in a 7B-sized model
   • The minimum VRAM requirements for evaluating models on a single GPU (I currently have access to 48GB GPU memory)

Thank you again for your dedication to the open source community. Your work is greatly appreciated!

Best regards