---
title: "Get started with External Integrations"
description: "Create a Hugging Face integration in a TIR project and see it in the list, ready to pick in a model endpoint or fine-tuning job."
content_type: "get-started"
product: "integrations"
platform: "TIR"
url: "/tir/integrations/get-started/"
---

# Get started with External Integrations

Connect your Hugging Face account to a TIR project, so the project can use models and datasets from
the Hub with your access token. The integration then appears on **Manage External Integrations**.

You need:

- A TIR project you have access to, and a role that lets you create integrations.
- A Hugging Face access token. To get one, open
  [Hugging Face Settings, Access Tokens](https://huggingface.co/settings/tokens).

## Create the integration

1. In TIR, open your project. In the side menu, under **PLATFORM SERVICES**, select **External
   Integrations**.
2. Select **Create Integration**.
3. On **Select Integration Type**, select **Hugging Face**.
4. In **Integration Name**, keep the suggested name or type your own. Use lowercase letters, digits, and
   hyphens, start with a letter, end with a letter or digit, and keep it to 30 characters or fewer.
5. In **Hugging Face Token**, paste your token. It must be free of spaces.
6. Select **Create**.

TIR checks the token with Hugging Face and saves the integration only once Hugging Face accepts it. You are
then back on the list with the new integration in it. If Hugging Face refuses the token, a message explains
why; to find out what each message means, see
[Troubleshooting External Integrations](/tir/integrations/troubleshooting/integrations/).

## Use it

Pick the integration in the form that asks for it, such as **HF Token Integration** when you create a model
endpoint from a Hugging Face model. See
[Where integrations are used](/tir/integrations/concepts/where-integrations-are-used/).

To connect GitHub or Weights & Biases instead, see
[Create a GitHub integration](/tir/integrations/how-to/create-a-github-integration/) and
[Create a Weights & Biases integration](/tir/integrations/how-to/create-a-weights-and-biases-integration/).