Anyone willing to collaborate on a eli5 guide?

[salamanders]

I got it running! The problem is,

For example,

Since CLIP offers a shared image/text feature space, and RDMs learn to cover a neighborhood of a given example during training, we can directly take a CLIP text embedding of a given prompt and condition on it. Run this mode via...

Say what? So these are finding midpoints between existing images? How is this different from the txt2img.py

Speaking of which, txt2img.py works great! But... what is ddim_eta? ddim_steps? should we ever NOT use plms?

I started writing down what little I know in https://docs.google.com/document/d/1C7CetthyMeIK9I41Icx7JZuir5IN8Ka5rXLa6f0B3zs/edit?usp=sharing but would welcome anything noobs like me discover!