Dataset Viewer
Auto-converted to Parquet Duplicate
action
stringclasses
23 values
source_filename
stringlengths
16
36
theme
stringclasses
2 values
size
stringclasses
4 values
badge
imagewidth (px)
164
500
Built with ML Intern
built-with-ml-intern-lg-dark.svg
dark
lg
Built with ML Intern
built-with-ml-intern-lg.svg
light
lg
Built with ML Intern
built-with-ml-intern-md-dark.svg
dark
md
Built with ML Intern
built-with-ml-intern-md.svg
light
md
Built with ML Intern
built-with-ml-intern-sm-dark.svg
dark
sm
Built with ML Intern
built-with-ml-intern-sm.svg
light
sm
Built with ML Intern
built-with-ml-intern-xl-dark.svg
dark
xl
Built with ML Intern
built-with-ml-intern-xl.svg
light
xl
Dataset on HF
dataset-on-hf-lg-dark.svg
dark
lg
Dataset on HF
dataset-on-hf-lg.svg
light
lg
Dataset on HF
dataset-on-hf-md-dark.svg
dark
md
Dataset on HF
dataset-on-hf-md.svg
light
md
Dataset on HF
dataset-on-hf-sm-dark.svg
dark
sm
Dataset on HF
dataset-on-hf-sm.svg
light
sm
Dataset on HF
dataset-on-hf-xl-dark.svg
dark
xl
Dataset on HF
dataset-on-hf-xl.svg
light
xl
Deploy on Spaces
deploy-on-spaces-lg-dark.svg
dark
lg
Deploy on Spaces
deploy-on-spaces-lg.svg
light
lg
Deploy on Spaces
deploy-on-spaces-md-dark.svg
dark
md
Deploy on Spaces
deploy-on-spaces-md.svg
light
md
Deploy on Spaces
deploy-on-spaces-sm-dark.svg
dark
sm
Deploy on Spaces
deploy-on-spaces-sm.svg
light
sm
Deploy on Spaces
deploy-on-spaces-xl-dark.svg
dark
xl
Deploy on Spaces
deploy-on-spaces-xl.svg
light
xl
Deploy on Spaces
deploy-to-spaces-lg.svg
light
lg
Deploy on Spaces
deploy-to-spaces-md.svg
light
md
Deploy on Spaces
deploy-to-spaces-sm.svg
light
sm
Duplicate this Space
duplicate-this-space-lg-dark.svg
dark
lg
Duplicate this Space
duplicate-this-space-lg.svg
light
lg
Duplicate this Space
duplicate-this-space-md-dark.svg
dark
md
Duplicate this Space
duplicate-this-space-md.svg
light
md
Duplicate this Space
duplicate-this-space-sm-dark.svg
dark
sm
Duplicate this Space
duplicate-this-space-sm.svg
light
sm
Duplicate this Space
duplicate-this-space-xl-dark.svg
dark
xl
Duplicate this Space
duplicate-this-space-xl.svg
light
xl
Follow me on HF
follow-me-on-HF-lg-dark.svg
dark
lg
Follow me on HF
follow-me-on-HF-lg.svg
light
lg
Follow me on HF
follow-me-on-HF-md-dark.svg
dark
md
Follow me on HF
follow-me-on-HF-md.svg
light
md
Follow me on HF
follow-me-on-HF-sm-dark.svg
dark
sm
Follow me on HF
follow-me-on-HF-sm.svg
light
sm
Follow me on HF
follow-me-on-HF-xl-dark.svg
dark
xl
Follow me on HF
follow-me-on-HF-xl.svg
light
xl
Follow us on HF
follow-us-on-hf-lg-dark.svg
dark
lg
Follow us on HF
follow-us-on-hf-lg.svg
light
lg
Follow us on HF
follow-us-on-hf-md-dark.svg
dark
md
Follow us on HF
follow-us-on-hf-md.svg
light
md
Follow us on HF
follow-us-on-hf-sm-dark.svg
dark
sm
Follow us on HF
follow-us-on-hf-sm.svg
light
sm
Follow us on HF
follow-us-on-hf-xl-dark.svg
dark
xl
Follow us on HF
follow-us-on-hf-xl.svg
light
xl
Build a demo
ml-intern-action-demo-dark.svg
dark
null
Build a demo
ml-intern-action-demo.svg
light
null
Evaluate it
ml-intern-action-evaluate-dark.svg
dark
null
Evaluate it
ml-intern-action-evaluate.svg
light
null
Fine-tune on your data
ml-intern-action-finetune-dark.svg
dark
null
Fine-tune on your data
ml-intern-action-finetune.svg
light
null
Quantize it
ml-intern-action-quantize-dark.svg
dark
null
Quantize it
ml-intern-action-quantize.svg
light
null
Reproduce it
ml-intern-action-reproduce-dark.svg
dark
null
Reproduce it
ml-intern-action-reproduce.svg
light
null
Train a model on it
ml-intern-action-train-dark.svg
dark
null
Train a model on it
ml-intern-action-train.svg
light
null
Model on HF
model-on-hf-lg-dark.svg
dark
lg
Model on HF
model-on-hf-lg.svg
light
lg
Model on HF
model-on-hf-md-dark.svg
dark
md
Model on HF
model-on-hf-md.svg
light
md
Model on HF
model-on-hf-sm-dark.svg
dark
sm
Model on HF
model-on-hf-sm.svg
light
sm
Model on HF
model-on-hf-xl-dark.svg
dark
xl
Model on HF
model-on-hf-xl.svg
light
xl
Open a Discussion
open-a-discussion-lg-dark.svg
dark
lg
Open a Discussion
open-a-discussion-lg.svg
light
lg
Open a Discussion
open-a-discussion-md-dark.svg
dark
md
Open a Discussion
open-a-discussion-md.svg
light
md
Open a Discussion
open-a-discussion-sm-dark.svg
dark
sm
Open a Discussion
open-a-discussion-sm.svg
light
sm
Open a Discussion
open-a-discussion-xl-dark.svg
dark
xl
Open a Discussion
open-a-discussion-xl.svg
light
xl
Open a Pull Request
open-a-pr-lg-dark.svg
dark
lg
Open a Pull Request
open-a-pr-lg-light.svg
light
lg
Open a Pull Request
open-a-pr-lg.svg
light
lg
Open a Pull Request
open-a-pr-md-dark.svg
dark
md
Open a Pull Request
open-a-pr-md.svg
light
md
Open a Pull Request
open-a-pr-sm-dark.svg
dark
sm
Open a Pull Request
open-a-pr-sm.svg
light
sm
Open a Pull Request
open-a-pr-xl-dark.svg
dark
xl
Open a Pull Request
open-a-pr-xl.svg
light
xl
Open in HF Spaces
open-in-hf-spaces-lg-dark.svg
dark
lg
Open in HF Spaces
open-in-hf-spaces-lg.svg
light
lg
Open in HF Spaces
open-in-hf-spaces-md-dark.svg
dark
md
Open in HF Spaces
open-in-hf-spaces-md.svg
light
md
Open in HF Spaces
open-in-hf-spaces-sm-dark.svg
dark
sm
Open in HF Spaces
open-in-hf-spaces-sm.svg
light
sm
Open in HF Spaces
open-in-hf-spaces-xl-dark.svg
dark
xl
Open in HF Spaces
open-in-hf-spaces-xl.svg
light
xl
Paper page
paper-page-lg-dark.svg
dark
lg
Paper page
paper-page-lg.svg
light
lg
Paper page
paper-page-md-dark.svg
dark
md
Paper page
paper-page-md.svg
light
md
End of preview. Expand in Data Studio

Hugging Face Badges in Lance

Search the official Hugging Face badges by action and retrieve the original artwork by filename. The collection includes badges for sharing a dataset, deploying an app on Spaces, opening a pull request, and using ML Intern. Original Hugging Face Badges dataset and pinned source revision

Data

Lance table Rows Contents
lance/assets.lance 149 All original files, including 146 SVG variants, the thumbnail, source card, and Git attributes. Each row stores the original filename, bytes, byte length, and SHA-256 hash. SVG rows include the theme and any source label, target link, or title.
lance/actions.lance 23 One row per badge action, with its label, the text embedded for search, all filenames for that action, and a retrieval vector.

The upstream catalog has no train, validation, or test splits. Data Studio displays a Parquet preview of all 146 SVG variants rendered as PNGs under its default train split. The native Lance assets retain the original SVG bytes.

Search for badge actions

The source provides no embeddings. This release embeds each badge action in English using its source README heading or SVG title and a short description of the intended action, stored in embedded_text. Light and dark themes, sizes, and alternate filenames for the same action share one search result. The vectors use the first 256 dimensions of the encoder's Matryoshka representation with L2 normalization. The example uses the same encoder for the query and reports cosine similarity, with higher scores indicating closer matches. Pinned static retrieval encoder revision

pip install pylance==14.0.3 huggingface_hub==2.4.0 sentence-transformers==6.3.0
import lance
from huggingface_hub import HfApi
from sentence_transformers import SentenceTransformer

repo = "lance-format/hf-badges-lance"
root = f"hf://datasets/{repo}"
revision = HfApi(token=False).dataset_info(repo, token=False).sha
storage_options = {"hf_revision": revision}
encoder = SentenceTransformer(
    "sentence-transformers/static-retrieval-mrl-en-v1",
    revision="f60985c706f192d45d218078e49e5a8b6f15283a",
    truncate_dim=256,
    device="cpu",
)
query = "Show a badge to deploy my application on Hugging Face Spaces"
query_vector = encoder.encode_query(query, normalize_embeddings=True)
actions = lance.dataset(f"{root}/lance/actions.lance", storage_options=storage_options)
matches = actions.to_table(
    columns=["action_id", "action", "variant_filenames", "_distance"],
    nearest={"column": "embedding", "q": query_vector, "k": 5, "metric": "cosine"},
).to_pylist()

for match in matches:
    score = 1 - match["_distance"]
    filenames = ", ".join(match["variant_filenames"])
    print(f"{score:.4f}  {match['action_id']}  {match['action']}  {filenames}")

To save a badge as an original SVG, select a filename from the matches and read its bytes from the assets table.

from pathlib import Path

filename = matches[0]["variant_filenames"][0]
assets = lance.dataset(f"{root}/lance/assets.lance", storage_options=storage_options)
asset, = assets.to_table(
    filter=f"source_path = '{filename}'",
    columns=["source_bytes"],
).to_pylist()
Path(filename).write_bytes(asset["source_bytes"])

License and attribution

Hugging Face published the original artwork and documentation under the MIT license. This release adds the Lance tables, badge-action search, and PNG previews. Original dataset card · MIT license

Downloads last month
18