Loading dataset

Description

Every model card on the Hugging Face Hub, joined with the Hub's own metadata and labelled, so you can find a model by what it is, what it needs and where it comes from. One row per model in the public model-card mirror of the Hub: 885,448 models from the librarian-bots/model_cards_with_metadata snapshot of 23 October 2026, joined to the daily hub-stats mirror for task, library, tags, all-time downloads, trending score and dates.

From the Hub's tags and metadata, without a model in the loop: licence, languages from the language tags, base model and its relation (fine-tune, quantized, adapter, merge), training datasets, arXiv ids, architecture family, parameter count and weight precision from the safetensors index, weight size in GB, which GPU class the weights fit, GGUF availability, and how many models on the Hub derive from each one.

From a rule where the Hub decides it, and from a small classification model reading the card where it does not: the model's kind (instruct, base, reasoning, code, embedding, reranker, classifier, vision-language, image generation, speech and audio, vision, other), whether the card is real, empty or a test, and the context window the card states. Each labelled column has a companion naming its source.

Search it in plain words: "multilingual embedding model", "qwen3 fine-tunes under 4B", "reasoning model with a 128k context window", "speech recognition for Hindi". Models whose repositories are gated have no card in the mirror and are not here.

Tables1

  • Table healthStatus unavailablehf_all_modelsTable overviewPublished rows885,448Columns48 rows
    Status unavailable

Sources

Hugging Face – The AI community building the future.We’re on a journey to advance and democratize artificial intelligence through open source and open science.7 endpoints

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Usage rights
Allowed by terms of service.
Collection
7 requests across 7 endpoints

Collected endpoints

Every public model repository on the Hugging Face Hub with its author, task, library, tags, downloads, likes, trending score, safetensors parameter counts and base-model lineage, from the daily hub-stats mirror./datasets/cfahlgren1/hub-stats/resolve/main/models.parquet1 request
Hugging Face model cards from file 1 of 6 of the public card mirror, with the mirror's task, library, downloads and dates, read by byte range. Skips the cards whose text holds a NUL byte, which the reader refuses./datasets/librarian-bots/model_cards_with_metadata/resolve/main/data/train-00000-of-00006.parquet1 request
Hugging Face model cards from file 2 of 6 of the public card mirror, with the mirror's task, library, downloads and dates, read by byte range. Skips the cards whose text holds a NUL byte, which the reader refuses./datasets/librarian-bots/model_cards_with_metadata/resolve/main/data/train-00001-of-00006.parquet1 request
Hugging Face model cards from file 3 of 6 of the public card mirror, with the mirror's task, library, downloads and dates, read by byte range. Skips the cards whose text holds a NUL byte, which the reader refuses./datasets/librarian-bots/model_cards_with_metadata/resolve/main/data/train-00002-of-00006.parquet1 request
Hugging Face model cards from file 4 of 6 of the public card mirror, with the mirror's task, library, downloads and dates, read by byte range. Skips the cards whose text holds a NUL byte, which the reader refuses./datasets/librarian-bots/model_cards_with_metadata/resolve/main/data/train-00003-of-00006.parquet1 request
Hugging Face model cards from file 5 of 6 of the public card mirror, with the mirror's task, library, downloads and dates, read by byte range. Skips the cards whose text holds a NUL byte, which the reader refuses./datasets/librarian-bots/model_cards_with_metadata/resolve/main/data/train-00004-of-00006.parquet1 request
Hugging Face model cards from file 6 of 6 of the public card mirror, with the mirror's task, library, downloads and dates, read by byte range. Skips the cards whose text holds a NUL byte, which the reader refuses./datasets/librarian-bots/model_cards_with_metadata/resolve/main/data/train-00005-of-00006.parquet1 request
ClefServed by Clef (asia-south2), Clef (asia-southeast1), Clef (us-central1) and Clef (europe-west4)6 endpoints
Usage rights
Licensed by contract.
Collection
6 requests across 6 endpoints

Collected endpoints

A small classification model reads each model card of file 1 of 6 and labels the model's kind, the card's quality and the context window it states, as probabilities per label.labels_01 request
A small classification model reads each model card of file 2 of 6 and labels the model's kind, the card's quality and the context window it states, as probabilities per label.labels_11 request
A small classification model reads each model card of file 3 of 6 and labels the model's kind, the card's quality and the context window it states, as probabilities per label.labels_21 request
A small classification model reads each model card of file 4 of 6 and labels the model's kind, the card's quality and the context window it states, as probabilities per label.labels_31 request
A small classification model reads each model card of file 5 of 6 and labels the model's kind, the card's quality and the context window it states, as probabilities per label.labels_41 request
A small classification model reads each model card of file 6 of 6 and labels the model's kind, the card's quality and the context window it states, as probabilities per label.labels_51 request

Details

Contents
1 table · 885,448 rows · 48 columns
Updated
Published
Version
v1
License
Not stated
Visibility
Public
Publisher
Mostly Right
Topics
hugging face · model cards · language models +5

Activity

Views545+545 in the last 30 days
Likes0+0 in the last 30 days
Uses37+37 in the last 30 days

Discover

Support

Talk to Mostly Right

Loading chat…