stsavian

Results 10 comments of stsavian

how do i train upsampling 128->512 on a custom dataset?

@nicolai256 did you manage to run these experiments? Could you report the results and memory requirement? That would help me, Thanks

Training generates images with full red output

Interesting, I opened a similar issue for the following repository https://github.com/cloneofsimo/minDiffusion/issues/4

After training the sample pictures get some weird color tints

I am also facing a similar problem!

After training the sample pictures get some weird color tints

Hello, I have found that by predicting the target (x_o), instead of the noise (epsilon), the phenomenon is dramatically reduced. Have you tried to set predict_x_start to true? Looking forward...

Imagenet-C Scores dependent on Python Version?

Thanks for replicating the results, I am also having troubles with this.

When sampling a model trained with class_cond=True, do we need to give it the imagenet grountruth labels?

Thanks @choROPeNt for your answer! Actually my question was for the sampling stage, At the beginning I thought that we needed to provide the labels solely during training, and that...

Does max_beta=0.999 in cosine schedule make any sense ?

@unrealwill sorry for my possibly trivial question, when you say: > `In my personal loose implementation I have decided to use the prediction of the noise (Ho-style) instead of the...

Does max_beta=0.999 in cosine schedule make any sense ?

@unrealwill thanks for your kind reply! Indeed I am having some problems getting the model to converge. I would like to better explain my experiments to you. With my settings,...

Does max_beta=0.999 in cosine schedule make any sense ?

@unrealwill thanks for your advice! I will make good use of it!

Quality of the generated images

@cloneofsimo thanks so much for your advice!