Are there any recipes or repositories for performing post-training on the DeepSeek-R1 model, not based on the distilled models?

[JiayiFu]

Hi everyone,
I want to run some SFT experiments on the 671B DeepSeek-R1 model. However, I couldn’t find any recipes in this repository or on Hugging Face.
Does anyone know if there are any recipes or repositories for performing post-training on the DeepSeek-R1 model?
Thx!

[alan008]

https://github.com/open-thoughts/open-thoughts