tianweiy/DMD2

Issues

• set model to train mode

  #21 opened in 13 hours
  4

• Convert 4-step unet into a checkpoint file

  #20 opened 10 days ago
  6

• When i train SDXL-4steps model, i get OOM error, i use a single A100.

  #19 opened 18 days ago
  2

• sdv15 train with torchrun successful, but sdxl train with fsdp failed.

  #18 opened 22 days ago
  0

• How about DMD2 on img2img task?

  #17 opened 22 days ago
  2

• How does it compare to ByteDance/Hyper-SD or sd-community/sdxl-flash ?

  #16 opened 22 days ago
  2

• Please tell me where is the data files: sdxl_vae_latents_laion_500k_lmdb? I can not find it.

  #15 opened 12 days ago
  1

• Current CUDA Device does not support bfloat16. Please switch dtype to float16.

  #14 opened 23 days ago
  0

• Is it possible to integrate with diffusers?

  #12 opened 24 days ago
  1

• request default -1 seed and negative prompt

  #11 opened 12 days ago
  2

• Is it possible to integrate with HiDiffusion?

  #10 opened a month ago
  6

• Why is uncorrelated noise used in backward_simulation?

  #9 opened 25 days ago
  1

• Request for test_folder.py in ImageNet-64x64 Experiments

  #8 opened a month ago
  2

• Why add the discriminator header on the fake denoiser and not on the real denoiser where the parameters are fixed?

  #7 opened a month ago
  3

• Possibility for 30fps image generation?

  #6 opened a month ago
  6

• a Lora version to add to any Community Models?

  #5 opened a month ago
  6

• demo dead 💀

  #3 opened a month ago
  1

• How to visualize Figure3

  #2 opened 24 days ago
  1

• Plans for inpainting model.

  #1 opened a month ago
  1