Logic for training a ParaGuide model. See train.sh for a training template.
In our work, we fine-tune the original SSD-LM diffusion model (https://huggingface.co/xhan77/ssdlm), with our modified approach/architecture.
To fine-tune our Enron Model, you will first need to download our checkpoint (See models/download.sh), and preprocess your data (See data/).
train.py contains our training code.