> For the complete documentation index, see [llms.txt](https://perspective-ai.gitbook.io/perspective-ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://perspective-ai.gitbook.io/perspective-ai/our-tech-stack/about-huggingface.md).

# About HuggingFace

**HuggingFace** is the world’s largest hub for open-source AI models. From cutting-edge large language models (LLMs) to specialized tools for classification, translation, image generation, and more — it’s where the global AI and machine learning community shares, collaborates, and builds.

With PerspectiveAI, you'll be able to access many of these models directly, instantly, and without needing to manage any additional infrastructure.

{% embed url="<https://youtu.be/ga5iutse9sk>" %}

***

### What is HuggingFace?

At its core, Hugging Face is a platform and ecosystem for:

* **Pretrained AI models**: Thousands of models across language processing LLMs, computer vision, audio, and multimodal tasks.
* **Datasets**: Shared, versioned datasets used for training, benchmarking, and experimentation.
* **Transformers library**: One of the most widely-used Python libraries for working with modern LLMs.
* **Spaces**: Lightweight hosted apps built with models, enabling interactive demos and collaborative tools.

It’s a central part of the open-source AI movement, allowing anyone — from researchers to hobbyists — to experiment with state-of-the-art models.

***

### How PerspectiveAI Uses HuggingFace

With our Hugging Face integration, you’re not just browsing models — you’re using them directly through a streamlined, unified interface.

#### What You Can Do:

* **Select and prompt any supported model** right inside PerspectiveAI, just like you would with GPT-4 or Claude.
* **Compare performance across providers** (e.g. Hugging Face, OpenAI, Google, Anthropic) in one interface.
* **Use supported HuggingFace models as part of a multi-model workflow**, without any additional setup.

No API keys. No switching tabs. No setup.&#x20;

***

### Coming Soon: HuggingFace + Tokenization

We’re extending our support for HuggingFace in powerful new ways:

#### Tokenize & Deploy - Developers

You’ll be able to:

* Upload and host your own HuggingFace model through PerspectiveAI,
* Tokenize it with **$POV,**
* Earn rewards when your model is used,

#### Stake on Models - Users

* Stake $POV tokens on HuggingFace models you believe in,
* Receive rewards based on actual usage,
* Help surface the most useful and high-quality models,

#### Fine-Tuning Support - Everyone

* Get rewarded for contributing data or feedback to improve model performance,
* Earn $POV based on the usage and effectiveness of those improvements,

This creates a direct feedback and reward loop for builders, contributors, and communities.

***

### Use Cases

* Quickly try out community-built models without code,
* Explore lesser-known or fine-tuned models for niche tasks,
* Prototype multi-model workflows using a mix of open and closed-source tools,
* Support the growth of open AI by staking, contributing, or publishing your own models,

***

### How to Get Started

1. **Log in to PerspectiveAI** at [app.perspectiveAI.xyz](https://app.perspectiveAI.xyz),
2. Click **New Chat,**
3. Select a HuggingFace model from the dropdown (when we release support for new models),
4. Prompt it, test it, compare it,
5. Stay tuned for tokenization, staking, and agent integration.

You now have access to thousands of open models — with none of the usual friction!