[Feature request] Dreambooth deepspeed

[2x-y]

According to this reddit post, Dreambooth deepspeed appears to reduce VRAM usage to a level that many more users will be able to use.

Would it be feasible to incorporate it into the UI?

[alexfredo]

Yeah that would be awesome ! textual inversion can run under 6gb too

[ClashSAN]

How would this work if we do not have the diffusers SD?

[EKAPPPAEKAPPA]

Yeah that would be awesome ! textual inversion can run under 6gb too

Any way of showing a discussion bout this????

[0xdevalias]

Potentially Related:

• Implement new paper: Dreambooth-StableDiffusion, Google Imagen based Textual Inversion alternative #914
• Running AUTOMATIC1111 / stable-diffusion-webui with Dreambooth fine-tuned models #1429
• [Feature Request]: Dreambooth on 8GB VRam GPU (holy grail) #3586
• DreamBooth training in under 8 GB VRAM and textual inversion under 6 GB! #1741
• Dreambooth #2002
  • Dreambooth #2002 (comment)

    • Closing, opening new PR to squash commits and make it clean.

• Dreambooth: Ready to go! #3995
  • Dreambooth: Ready to go! #3995 (comment)

    • Please give this a look and merge.

[0xdevalias]

Dreambooth is available as an extension now, see below:

Closing this, as I've now started a repo with a standalone extension based on ShivShiram's repo here:

https://github.com/d8ahazard/sd_dreambooth_extension

Please feel free to test and yell at me there. I've added requirements installer, multiple concept training via JSON, and moved some bit about.

UI still needs fixing, some stuff broken there, but it should be able to train a model for now.

Originally posted by @d8ahazard in #3995 (comment)

[mezotaken]

https://github.com/d8ahazard/sd_dreambooth_extension models and training are supported via the extension, so i'm closing multiple DB issues, as no related changes could happen in this repo. Looks like minimal requirement is 10GB after all. Plus, it supports LORA now, which supposedly should work on 8GB.