Open libai-lab opened 2 weeks ago
Theoretically any diffusion inference acceleration work can be used, we don't have enough time to do acceleration for inference. If you can help, we'd really appreciate it!
Theoretically any diffusion inference acceleration work can be used, we don't have enough time to do acceleration for inference. If you can help, we'd really appreciate it!
there is a very Simple examples
Is the lambdalabs/sd image variations diffusers model sd1.4? Is it possible to use the SD1.5 model to generate acceleration through a few steps of hypersd, or is there any other solution that can optimize acceleration? I see that using the current model requires 50 steps. Can we use other acceleration schemes for single or multi-step image generation