n8n Workflow: Use an Open-Source LLM via Hugging Face

This n8n workflow demonstrates how to integrate an open-source Large Language Model (LLM) hosted on Hugging Face into your n8n workflows. It allows you to receive a chat message, process it using a Hugging Face Inference Model, and potentially build a more complex LangChain.

What it does

This workflow simplifies the process of interacting with open-source LLMs by:

1. Listening for chat messages: It acts as a trigger, initiating the workflow whenever a new chat message is received.
2. Initializing a Hugging Face Inference Model: It sets up a connection to a specified Hugging Face Inference endpoint, allowing you to leverage various open-source LLMs.
3. Processing with a Basic LLM Chain: It uses a LangChain "Basic LLM Chain" node, which acts as an orchestrator to send the incoming chat message to the Hugging Face model and receive its response.

Prerequisites/Requirements

To use this workflow, you will need:

• n8n Instance: A running n8n instance.
• Hugging Face Account (Optional, but recommended for custom models/endpoints): While you can use public Hugging Face Inference API endpoints, having an account allows for more flexibility, including using private models or dedicated inference endpoints.
• Hugging Face API Token (Optional): If you are using private models or exceed rate limits on public endpoints, you might need a Hugging Face API token configured as an n8n credential.

Setup/Usage

1. Import the workflow:
   • Copy the provided JSON code.
   • In your n8n instance, go to "Workflows" and click "New".
   • Click the "Import from JSON" button and paste the copied JSON.
   • Click "Import".
2. Configure the Hugging Face Inference Model node:
   • Locate the "Hugging Face Inference Model" node.
   • Click on it to open its settings.
   • You will need to specify the Model ID of the Hugging Face model you wish to use (e.g., google/flan-t5-xxl).
   • If you need to use a Hugging Face API Token, ensure you configure an Hugging Face Credential in n8n and select it in this node's settings.
3. Configure the Chat Trigger:
   • The "When chat message received" node is a trigger. You might need to integrate it with a specific chat platform (e.g., Slack, Discord, Telegram) if you want to receive messages from external sources. For testing, you can manually execute the workflow and provide a dummy chat message.
4. Activate the workflow:
   • After configuring the nodes, click the "Activate" toggle in the top right corner of the workflow editor to enable it.

Once activated, the workflow will listen for incoming chat messages, send them to the configured Hugging Face LLM, and the LLM's response will be available in the output of the "Basic LLM Chain" node for further processing in your workflow.