datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
olmoearth-paper-embeddings
OlmoEarth — Foundation-Model Embeddings for Paper Table 2
This dataset contains pre-extracted embeddings from 26 Earth-observation
foundation models evaluated on the 24 downstream tasks that make up
Table 2 of the OlmoEarth paper:
OlmoEarth: Stable Latent Image Modeling for Multimodal Earth Observation
AI2, 2025. arXiv:2511.13655.
For every supported (model, task) pair we ran the model's encoder over the
task's train / validation / test splits with the paper-best… See the full description on the dataset page: https://huggingface.co/datasets/allenai/olmoearth-paper-embeddings.Danbooru-WD-EVA-EmbeddingsThis dataset includes WD EVA v2 large embeddings for danbooru images. Pixiv will be added later. Tensors under w are direct outputs and match indexes for WD EVA model. Tensors under e are from WD EVA as well however these strips the classifiaction head, they are smaller and suitable for deduplication computing for example.
The indexes of w, e and f (filename) match.
WD EVA Model: https://huggingface.co/SmilingWolf/wd-eva02-large-tagger-v3
relaion2b-natural-embeddings
LAION-Natural Embeddings: CLIP ViT-H/14 Features for ~500M Natural Photographs (CCN 2025, Roth & Hebart)
LAION-Natural Embeddings provides pre-computed CLIP ViT-H/14 embeddings for ~500 million natural photographs from ReLAION-2B, filtered using the LAION-Natural naturalness classifier (score > 0.7).
Also known as: LAION-Natural Embeddings · ReLAION-Natural Embeddings · LAION-2B-Natural Embeddings
Part of the LAION-Natural dataset family, introduced in: How to sample the… See the full description on the dataset page: https://huggingface.co/datasets/andropar/relaion2b-natural-embeddings.road-images-and-embeddings
Norwegian Road Images with Embeddings (Trondheim Area)
A dataset of 34,908 road images from the Trondheim region of Norway (~40km radius), captured by Statens vegvesen (Norwegian Public Roads Administration) in 2025. Each image is paired with rich geospatial metadata, nearest address information, and a 3072-dimensional image embedding from Google's gemini-embedding-2-preview model.
Dataset Structure
Each example contains:
Field
Type
Description
image
Image… See the full description on the dataset page: https://huggingface.co/datasets/thomasht86/road-images-and-embeddings.DinoBloom_hemato_embeddings
DinoBloom Hemato Patient Embeddings
Per-patient image embeddings of peripheral blood smears, extracted with the DinoBloom-B foundation model (code, paper).
Each .h5 file contains the stacked DinoBloom-B embeddings for all single-cell crops of one patient.
Contents
patient_embeddings/
├── caitomorph/ # 409 patients — caitomorph cohort (Dasdelen et al., 2026)
├── aml_hehr/ # 189 patients — AML genetic-subtype cohort (Hehr et al., 2023)
└── apl_aml/ # 106… See the full description on the dataset page: https://huggingface.co/datasets/MarrLab/DinoBloom_hemato_embeddings.local-embeddings-2022
Local Embeddings Dataset
Multi-temporal satellite imagery dataset for phenology embedding training.
Dataset Description
This dataset contains multi-spectral satellite tiles across 6 months (April-September 2022) with 16 bands per tile.
Dataset Structure
local_embeddings/
├── alphaearth_embeddings_tiles_202204/ (263 tiles)
├── alphaearth_embeddings_tiles_202205/ (263 tiles)
├── alphaearth_embeddings_tiles_202206/ (263 tiles)
├──… See the full description on the dataset page: https://huggingface.co/datasets/gabrielireland/local-embeddings-2022.spheer-fm-embeddings
Spheer FM Embeddings
Annual, 10 m, per-pixel embeddings pre-computed with Spheer FM Albatross, a self-supervised geospatial foundation model trained on Sentinel-2 time series.
Spheer FM is deliberately specialised: it is trained on European Sentinel-2 time series, with a focus on nature and biodiversity. Unlike foundation models built primarily around spatial image structure, Spheer FM places temporal land-surface dynamics at the centre of its representation. Its temporal… See the full description on the dataset page: https://huggingface.co/datasets/spheer/spheer-fm-embeddings.merged_remote_landscapes_v1
Dataset Card for Merged Remote Landscapes dataset
Dataset summary
This is a merged version of following datasets:
torchgeo/ucmerced
NWPU-RESISC45
from datasets import load_dataset
dataset = load_dataset('EmbeddingStudio/merged_remote_landscapes_v1')
Categories
This is a union of categories from original datasets:
agricultural, airplane, airport, baseball diamond, basketball court, beach, bridge, buildings, chaparral, church, circular farmland, cloud… See the full description on the dataset page: https://huggingface.co/datasets/EmbeddingStudio/merged_remote_landscapes_v1.anima-style-embedding-500k-full-face
Anima Style Embedding Full-Frame and Face Corpus
Dataset summary
The corpus contains 500,000 person-bearing illustrations for open-set anime style-embedding research. Human artwork and Anima-generated images use independent style identities because an artist's original work and Anima's response to the corresponding artist tag are not the same visual distribution.
Source directory
Style identities
Images per identity
Images
synthetic/
5,000
50
250,000… See the full description on the dataset page: https://huggingface.co/datasets/ij/anima-style-embedding-500k-full-face.imagenet-1k-224-clip-embeddings
ImageNet-1k-224 CLIP Embeddings
Pre-computed CLIP image embeddings for every image in
mlnomad/imagenet-1k-224.
Columns
Column
Type
Description
original_index
int
Row index in the source dataset for cross-referencing
label
int (0–999)
ImageNet class index
embedding
List[float]
L2-normalised CLIP image embedding (768D)
Stats
Source: mlnomad/imagenet-1k-224 (train split)
Total images: 1281167
Embedding dim: 768
CLIP model:… See the full description on the dataset page: https://huggingface.co/datasets/mlnomad/imagenet-1k-224-clip-embeddings.vlm-compositionality-embeddings
VLM Compositionality Embeddings
Pre-computed image and text embeddings for the thesis "From Euclidean to Hyperbolic Vision-Language Spaces: A Study of Attribute–Object Compositionality" by Meelad Dashti (Politecnico di Torino & University of Twente, 2026).
Code repository: github.com/MelDashti/hyperbolic-vlm-compositionality
Models
Model
Geometry
Architecture
Training Data
CLIP ViT-L/14
Spherical
ViT-L/14
WIT (400M+ pairs)
DINOv2 ViT-L/14
Spherical
ViT-L/14… See the full description on the dataset page: https://huggingface.co/datasets/Meldashti/vlm-compositionality-embeddings.testing_qwen3vl_embeddings
Dataset Card for random_short_videos
This is a FiftyOne dataset with 412 samples.
Installation
If you haven't already, install FiftyOne:
pip install -U fiftyone
Usage
import fiftyone as fo
from fiftyone.utils.huggingface import load_from_hub
# Load the dataset
# Note: other available arguments include 'max_samples', etc
dataset = load_from_hub("harpreetsahota/testing_qwen3vl_embeddings")
# Launch the App
session = fo.launch_app(dataset)… See the full description on the dataset page: https://huggingface.co/datasets/harpreetsahota/testing_qwen3vl_embeddings.flux-classification-embeddingsThe dataset contains over 200 embeddings and labels for FLUX classification. The dataset should be used in conjunction with the embedding model.
embeddings.npy - Contains the embeddings
labels.npy - contains the labels.
danbooru-convnext-embeddings2
Dataset Card for Danbooru ConvNeXt Embeddings 2
Danbooru ConvNeXt 向量数据集 2
Dataset Details / 数据集详情
Dataset Description / 数据集描述
English:
This dataset contains approximately 5,312,000 image embeddings (vectors). It was generated by extracting features from the massive Danbooru anime image dataset using the convnext_large.dinov3_lvd1689m computer vision model. These embeddings represent the visual features of the images in a high-dimensional space… See the full description on the dataset page: https://huggingface.co/datasets/telecomadm1145/danbooru-convnext-embeddings2.fashion_embeddings