Could you please share with us the prompts (for the evaluation dataset) and evaluation code used to compute the PPL, PDV and FID metrics? During testing, I noticed that the direct use of LPIPS (perceptual loss) leads to results that do not match those in Table 1 of your paper.
This would help us a lot to ensure consistency in our comparison.
Hi!
Could you please share with us the prompts (for the evaluation dataset) and evaluation code used to compute the PPL, PDV and FID metrics? During testing, I noticed that the direct use of LPIPS (perceptual loss) leads to results that do not match those in Table 1 of your paper.
This would help us a lot to ensure consistency in our comparison.