I taught an LLM to optimize proteins. It proposed a better carbon capture enzyme. Introducing Pro-1, an 8b param reasoning model trained using GRPO towards a physics based reward function for protein stability. It takes in a protein sequence + text description + previous experimental... See more