Comment by leopoldj

The model supports both instruction and RL fine-tuning. It looks like people have started training it already [1]. The RL tuning may be only available to the Cloudflare customers, I couldn't be sure.

1. https://huggingface.co/solanaclawd/clef-solana-research