Benchmark against TEXTurePaper
Closed this issue · 2 comments
yosun commented
Hey I am curious how this performs (and which hardware you are using) for a mix of different models, and also against TEXTurePaper
reynoldscem commented
It looks like the TEXTure paper is very similar concurrent work. I'd be interested to see a comparison.
daveredrum commented
Metric-wise our method is bit higher than TEXTure (FID: 35.68 vs 39.09, KID (x10^-3): 7.74 vs 9.97).