Question:Meaning of gen_top_k, gen_top_p, temperature, and cond_scale

[mallorbc]

In the context of NLP models, I understand what top k top p, and temperature do with regard to sampling. Could someone explain what they mean(and cond_scale) with regards to a diffusion model?

If it was autoregressive, each token could be a color representation and these sampling parameters could change how the next token is selected. But for diffusion I am lost and would appreciate any insight.