varshhhy7/esm2-rl-designer

Research framework for fine-tuning a pretrained protein language model (ESM-2) using reinforcement learning. Uses LoRA and multi-objective rewards to steer protein generation toward stability, diversity, and constraint satisfaction, with experiments designed to study trade-offs under different reward designs.

语言：Jupyter Notebook · GitHub 星标：17 · Fork：1

在 GitHub 上查看原仓库