search

Sherrylone / PQDiff

[ICLR 2024] Continuous-Multiple Image Outpainting in One-Step via Positional Query and A Diffusion-based Approach Link: https://arxiv.org/abs/2401.15652
71 stars 1 forks source link

Training results #14

Closed bhosalems closed 1 month ago

bhosalems commented 4 months ago

I wanted to reproduce the results on flicker/sceneray dataset. After training 15000 steps i still see results that do not look good, any advice?

image

Sherrylone commented 3 months ago

You can increase the number of iterations to 100k.

ustczhouyu commented 3 months ago

I wanted to reproduce the results on flicker/sceneray dataset. After training 15000 steps i still see results that do not look good, any advice?

image

I met the same problem when I sample at 10000 steps, have you solved this problem?

mondalanindya commented 2 months ago

Decrease your learning rate, worked for me!