unditional DDIM in fig 2

[Jun]

Dear all,

Thanks for sharing the awesome work!

Would it be possible for you to share a jupyter-notebook to reproduce the unconditional DDIM results in Fig 2?

[bram-w]

Leaving this open for now since I can put in a more formal version of this code but for those wondering about this here's a way to do it

[bram-w]

Commited the recon function (not tested but should be fine, re-open if issue) 6d70444

[zzzzzx-1115]

@bram-w Hello, I have the same problem about the 3th and 4th column of Fig 2: how do you perform deterministic noising to get the noised $x_T$ which will be used as the starting point of the denoising process? Or say, during this noising process, you use $\varepsilon_{uncondition}$, or $\varepsilon_{condition}$, or $\varepsilon_{uncondition} + w(\varepsilon_{condition}-\varepsilon_{uncondition})$?