Open yaseryacoob opened 2 years ago
Hi Yaser,
You can use the lines of code here https://github.com/AlexiaJM/score_sde_fast_sampling/blob/5da8f3fe103ee5ac3c3a336f16cc06c9541f0ed9/experiments.sh#L165 to compare the various different sampling methods on FFHQ-256. You'll want to replace ffhq_256.py with ffhq.py in the lines of code.
Based on the paper, for optimal results, you'll want to use this line: https://github.com/AlexiaJM/score_sde_fast_sampling/blob/5da8f3fe103ee5ac3c3a336f16cc06c9541f0ed9/experiments.sh#L170
You can download FFHQ from https://github.com/NVlabs/ffhq-dataset.
Alexia, thanks for the quick response. Let's clarify
thanks for further clarifications.
Thanks Alexia, I tried this route, but the ml-collections and absl seem to have some issues beyond my knowledge. I will let go of it for now and revisit it in the future.
Thanks for sharing your code, I am trying to compare the performance of your code to Songs on the FFHQ1024x1024 data. You seem to have a config file ffhq.py that might do that. I am not sure how to test this, I also see the need for ffhq-r10.tfrecords and not sure on the checkpoint and options for the call.
Can you please provide suggestions? thanks