datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Stable-Diffusion-Prompts
Stable Diffusion Dataset
This is a set of about 80,000 prompts filtered and extracted from the image finder for Stable Diffusion: "Lexica.art". It was a little difficult to extract the data, since the search engine still doesn't have a public API without being protected by cloudflare.
If you want to test the model with a demo, you can go to: "spaces/Gustavosta/MagicPrompt-Stable-Diffusion".
If you want to see the model, go to: "Gustavosta/MagicPrompt-Stable-Diffusion".
stable-diffusion-v1-5-glazed
Dataset Card for Stable Diffusion v1.5 Glazed Samples
Dataset Description
Dataset Summary
This dataset contains image samples originally generated by runwayml/stable-diffusion-v1-5
and subsequently processed by Glaze tool.
Supported Tasks and Leaderboards
[More Information Needed]
Languages
[More Information Needed]
Dataset Structure
Data Instances
[More Information Needed]
Data Fields
[More Information… See the full description on the dataset page: https://huggingface.co/datasets/hanamizuki-ai/stable-diffusion-v1-5-glazed.diffusion-pretrain-set-ft1
diffusion-pretrain-set-ft1
A multi-source image-caption pretraining dataset assembled from ten upstream
sources via a uniform ingest pipeline. Designed for a full pretrain or finetune
pipeline meant to curate for any major diffusion model preliminary, with the sole
intent to create a more powerful baseline preliminary train and a baseline
for synthesizing images to train the next generation of the VLM model.
This is a lot like the snake eating it's own tail, so it must be… See the full description on the dataset page: https://huggingface.co/datasets/AbstractPhil/diffusion-pretrain-set-ft1.Diffusion4D-Animated-Raw
Diffusion4D Animated Assets
This dataset provides animated 3D assets referenced by Diffusion4D and
Objaverse-XL in a directly browsable format. The default split contains 67,988
rows. Each row includes metadata, a preview image, and a short preview video so
that assets can be inspected in the Hugging Face Data Studio without first
downloading the original 3D file.
The repository also mirrors available raw assets and keeps their original
source links and hashes. The current… See the full description on the dataset page: https://huggingface.co/datasets/DenisKochetov/Diffusion4D-Animated-Raw.Stable_Diffusion_3_RecaptionThis dataset is the one specified in the stable diffusion 3 paper which is composed of the ImageNet dataset and the CC12M dataset.
I used the ImageNet 2012 train/val data and captioned it as specified in the paper: "a photo of a 〈class name〉" (note all ids are 999,999,999)
CC12M is a dataset with 12 million images created in 2021. Unfortunately the downloader provided by Google has many broken links and the download takes forever.
However, some people in the community publicized the dataset.… See the full description on the dataset page: https://huggingface.co/datasets/gmongaras/Stable_Diffusion_3_Recaption.diffusion_db_dedupe_from50k_train
Dataset Card for "diffusion_db_dedupe_from50k_train"
More Information needed
diffusion-pretrain-set-ft1-1024
diffusion-pretrain-set-ft1-1024
1024px (2x) upscale of AbstractPhil/diffusion-pretrain-set-ft1.
WARNING
MUCH OF THIS DATA WAS MODEL UPSCALED USING RAPID UPSCALERS.
THIS IS NOT CONSISTENTLY HIGH FIDELITY NOR IS IT EVEN CLOSE TO FAIR FIDELITY AT TIMES.
PLEASE use this ONLY for pretraining, new concepts, and simple design purposes ONLY. HEAVILY PRUNE FOR FINETUNING.
Thank you, good luck my friends.
Details
Model: realesr-general-x4v3 (SRVGG Compact… See the full description on the dataset page: https://huggingface.co/datasets/AbstractPhil/diffusion-pretrain-set-ft1-1024.DiffusionDream_DatasetThis is the dataset of the diffusion dream dataset. The dataset contains the following columns:
info: A string describing the action taken in the frame
keyword: A string describing the keyword of the action
action: A string describing the action taken in the frame
current_frame: The current frame of the video
previous_frame_1: The frame before the current frame
previous_frame_2: The frame before the previous frame
previous_frame_3: The frame before the previous frame
previous_frame_4: The… See the full description on the dataset page: https://huggingface.co/datasets/fffffchopin/DiffusionDream_Dataset.diffusionprint_dataset
DiffusionPrint Patch Dataset
A dataset of 64x64 image patches for contrastive learning of diffusion-based inpainting forensics.
Each patch comes from either a real image or an AI-inpainted region generated by one of three diffusion models.
Columns
Column
Type
Description
image
bytes (PNG)
Lossless 64x64 RGB patch
master_index
int64
Row index in the original memmap archive
patch_path
string
Original relative path of the patch
category
string
real… See the full description on the dataset page: https://huggingface.co/datasets/giakoupg/diffusionprint_dataset.stable-diffusion-prompts-stats-full-uncensoredlayout_diffusion_hypersimThis repository contains the data for SceneCraft: Layout-Guided 3D Scene Generation.
Project page: https://orangesodahub.github.io/SceneCraft
Code: https://github.com/OrangeSodahub/SceneCraft
diffusiondb_2m_random_50k
Dataset Card for "diffusiondb_2m_random_50k"
More Information needed
person-centric-images-stable-diffusion-v1-1lexica-stable-diffusion-v1-5
Stable Diffusion Dataset
This is a set of about 80,000 Image-Prompt pairs generated by stable-diffusion-v1-5.
The Prompts come from dataset Stable-Diffusion-Prompts which filtered and extracted from the image finder for Stable Diffusion: "Lexica.art".
1m2r-real-diffusion2This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "kinova-arx",
"total_episodes": 72,
"total_frames": 32350,
"total_tasks": 1,
"total_videos": 0,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 10,
"splits": {
"train": "0:72"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/rdoshi21/1m2r-real-diffusion2.audio-diffusion-1024Over 20,000 256x256 mel spectrograms of 5 second samples of music from my Spotify liked playlist. The code to convert from audio to spectrogram and vice versa can be found in https://github.com/teticio/audio-diffusion along with scripts to train and run inference using De-noising Diffusion Probabilistic Models.
x_res = 1024
y_res = 1024
sample_rate = 44100
n_fft = 2048
hop_length = 512
diffusionpen-hebrew-handwriting
DiffusionPen Hebrew Handwriting
A large synthetic dataset of Hebrew handwritten text lines with ground-truth
transcriptions, for training and evaluating handwritten text recognition (HTR / OCR)
models. Every image is a single line of right-to-left Hebrew handwriting synthesized by
DiffusionPen — a style-conditioned latent-diffusion handwriting generator — in one of
491 distinct writer styles, and quality-filtered by an independent OCR pass.
149,952 line images, 491 writer… See the full description on the dataset page: https://huggingface.co/datasets/cyttic/diffusionpen-hebrew-handwriting.civitai-stable-diffusion-337k
How to Use
from datasets import load_dataset
dataset = load_dataset("thefcraft/civitai-stable-diffusion-337k")
print(dataset['train'][0])
download images
download zip files from images dir
https://huggingface.co/datasets/thefcraft/civitai-stable-diffusion-337k/tree/main/images
it contains some images with id
from zipfile import ZipFile
with ZipFile("filename.zip", 'r') as zObject: zObject.extractall()
Dataset Summary
GitHub URL:-… See the full description on the dataset page: https://huggingface.co/datasets/thefcraft/civitai-stable-diffusion-337k.DiffusionJockey
Dataset Card for "DiffusionJockey"
More Information needed
Benchmark-Images-for-Stable-Diffusion-Biasbm-subnet-stable-diffusion-xl-base-1.0diffusion_db_dedup_from50k_train_v2
Dataset Card for "diffusion_db_dedup_from50k_train_v2"
More Information needed
text-to-image-diffusiondb-2M
DiffusionDB text-to-image subset
A cleaned, safety-filtered image-prompt dataset for training a text-to-image
model, built from DiffusionDB.
Built on Hugging Face Jobs directly from poloclub/diffusiondb. It covers
part_id 1-20 (20,000 source images) before filtering. The same content is
also kept on the 20k-subset branch.
Load it with:
load_dataset("whosouravsharma/text-to-image-diffusiondb-2M")
Note on the repo name: despite "2M" in the name, this is a small slice of… See the full description on the dataset page: https://huggingface.co/datasets/whosouravsharma/text-to-image-diffusiondb-2M.diffusiondb_random_10k_zh_v1
Dataset Card for "diffusiondb_random_10k_zh_v1"
svjack/diffusiondb_random_10k_zh_v1 is a dataset that random sample 10k English samples from diffusiondb and use NMT translate them into Chinese with some corrections.
it used to train stable diffusion models in svjack/Stable-Diffusion-FineTuned-zh-v0
svjack/Stable-Diffusion-FineTuned-zh-v1
svjack/Stable-Diffusion-FineTuned-zh-v2
And is the data support of https://github.com/svjack/Stable-Diffusion-Chinese-Extend which is a fine tune… See the full description on the dataset page: https://huggingface.co/datasets/svjack/diffusiondb_random_10k_zh_v1.diffusiondb_random_10k
Dataset Card for "diffusiondb_random_10k"
More Information needed
Dungeons-and-DiffusionThis is the dataset! Not the .ckpt trained model - the model is located here: https://huggingface.co/0xJustin/Dungeons-and-Diffusion/tree/main
The newest version has manually captioned races and classes, and the model is trained with EveryDream. 30 images each of: aarakocra, aasimar, air_genasi, centaur, dragonborn, drow,
dwarf, earth_genasi, elf, firbolg, fire_genasi, gith, gnome, goblin, goliath, halfling, human, illithid, kenku, kobold, lizardfolk, minotaur, orc, tabaxi, thrikreen, tiefling… See the full description on the dataset page: https://huggingface.co/datasets/0xJustin/Dungeons-and-Diffusion.audio-diffusion-512Over 20,000 512x512 mel spectrograms of 5 second samples of music from my Spotify liked playlist. The code to convert from audio to spectrogram and vice versa can be found in https://github.com/teticio/audio-diffusion along with scripts to train and run inference using De-noising Diffusion Probabilistic Models.
x_res = 512
y_res = 512
sample_rate = 22050
n_fft = 2048
hop_length = 512
diffusion_sftdiffusionpen-hebrew-handwriting-cer0
DiffusionPen Hebrew Handwriting — CER=0 clean subset
The highest-fidelity slice of DiffusionPen Hebrew Handwriting:
only the 33,082 line images that an independent Hebrew TrOCR read back with an exact
match (character error rate = 0.0). Built to test whether a smaller, label-clean set
trains a better recognizer than the full (noisier) 150k set.
33,082 line images, 491 writer styles
Splits (writer-independent, style-disjoint): train 26,607 / validation 3,187 / test 3,288
Every… See the full description on the dataset page: https://huggingface.co/datasets/cyttic/diffusionpen-hebrew-handwriting-cer0.wikipedia-bloom
Dataset Card for "wikipedia-bloom"
More Information needed