metadata

license: mit
task_categories:
  - image-classification
  - zero-shot-image-classification
  - text-to-image
language:
  - en
  - ja
tags:
  - art
  - anime
  - not-for-all-audiences
size_categories:
  - 1M<n<10M
annotations_creators:
  - no-annotation
source_datasets:
  - danbooru

E621 2024 SDXL VAE latents in 1k tar

Dedicated dataset to align both NebulaeWis/e621-2024-webp-4Mpixel and deepghs/e621_newest-webp-4Mpixel. "4MP-Focus" for average raw image resolution.
Latents are ARB with maximum size of 1024x1024 as the recommended setting in kohyas. Major reason is to make sure I can finetune with RTX 3090. VRAM usage will raise drastically after 1024.
Generated from prepare_buckets_latents_v2.py, modified from prepare_buckets_latents.py.
Used for kohya-ss/sd-scripts. In theory it may replace *.webp and *.txt along with meta_lat.json. Raw data is no longer required.
It took me around 10 days with 4x RTX 3090 to generate (with many PSU trips and I/O deadlocks). Perfect case would be 3-4 days only (18 it/s).
Download along with webp / txt, and then extract them all to single directory, and then you are good to go. Tags available in 6DammK9/e621_2024-tags-1ktar.
I still don't know how to work with multigpu trainning in Windows. Ultimately I may need to switch trainer. Use this repo if you are working well already.
The used VAE: madebyollin/sdxl-vae-fp16-fix
Most logs are lost. Verify with verify_npz.py. It should take 40 minutes (2000 it/s) if OS is super stable and you have a nice U.2 (Intel P4510 4T) and CPU (Intel Xeon 8358).

Extra: 12.5M Merged dataset for both danbooru and e621

Check out the meta_lat_merged.tar.gz. It is 23.8GB when decompressed.
The keys are casted in such pattern:

#250225: Relative to --train_data_dir="/tmp/dataset"
FOLDER_A = "danbooru/"
FOLDER_B = "e621/"

merged = {}

def cast_a(k):
    return f"{FOLDER_A}{k}"

def cast_b(k):
    return f"{FOLDER_B}{k}"

One of the best apporach is create a nested folder like /tmp/dataset/danbooru and /tmp/dataset/e621. Kohyas (torch.data.DataLoader) will support localized path.