# Overview

### What is Model Hub?

**Model Hub** is basically a **central repository of AI models**.

* You can **browse, download, share, and version-control** models.
* Provide tools for training, fine-tuning, and deploying models.

### How does it work?

#### Step 1. Pick a Base Model from the Model Catalog <a href="#contentify_0" id="contentify_0"></a>

* Search the model (e.g., `"Llama-3.2-1B"` for NLP, `"Qwen2-VL-72B"` for vision).
* These models already learned general representations.

#### Step 2. Launch Fine-tuning <a href="#contentify_1" id="contentify_1"></a>

* Use selected model to start fine-tuning

#### Step 3. Push Fine-tuned Model Back to the Hub <a href="#contentify_2" id="contentify_2"></a>

* After training, you upload your fine-tuned model to the hub.
* It gets a new version ID and can be shared with collaborators (or kept private).

#### Step 4. Use Anywhere <a href="#contentify_3" id="contentify_3"></a>

* Once fine-tuned, you can **download the model by SDK**.
* This ensures reproducibility and easy deployment.

### Why Model Hub?

1. **Centralized Access to Models**

* Without a hub, everyone would have to **hunt for models on GitHub repos, random blogs, or papers**.
* **Model Hub** is a **single catalog** where you can:
  * Search models by task (`text-classification`, `speech-to-text`, `image-segmentation`).
  * Compare architectures and benchmarks.
  * Reuse models with just one line of code.

2. **Reusability & Efficiency**

* Training large models from scratch is **expensive**.
* Hubs let you **reuse pretrained checkpoints**, so you only need to fine-tune.

3. **Collaboration & Sharing**

* Teams can **push fine-tuned models** to a hub → other team members can pull them instantly.
* This works for code: version control, forks, and community contributions.

4. **Deployment Ready**

* That means once your model is in the hub, you can:
  * Deploy it on cloud infrastructure.
  * Use it via REST APIs.
  * Scale it without managing servers.

5. **Governance & Version Control**

* Hubs track **different versions** of models.
* You know exactly which checkpoint was used in production (important for **MLOps & audits**).
* You can mark models as **public, private, or restricted**.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://ai-docs.fptcloud.com/fpt-ai-studio/services/model-hub/overview.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.