Comment on Efficient high-resolution image synthesis with linear diffusion transformerparentComments−cube22221yYeah I’d likely just pick the best scoring one (that is, the pick is made by the evaluation tool, not the model) - to simulate “whatever the receiver deemed best for what they wanted”.
Comments
Yeah I’d likely just pick the best scoring one (that is, the pick is made by the evaluation tool, not the model) - to simulate “whatever the receiver deemed best for what they wanted”.