OLLAMA + OPEN-WEBUI + PIPELINES + LANGFUSE

Introduction

This repository provides a setup for integrating OLLAMA, OPEN-WEBUI, PIPELINES, and LANGFUSE using Docker. Follow the steps below to get everything up and running.

Prerequisites

• Docker and required GPU drivers installed on your system.

Installation

1. Clone this repository:

   git clone https://github.com/karaketir16/openwebui-langfuse.git
   cd openwebui-langfuse

2. Run the setup script:

   ./run-compose.sh

   or

   docker compose -f docker-compose.yaml -f langfuse-v3.yaml up -d
   # default driver is nvidia

Configuration

Langfuse Setup

0. Documentation

   • You can find up-to-date documentation here.

1. Download the langfuse_filter_pipeline.py file (only if offline):

   • If your setup does not have internet access:
     • You can manually download the script from: https://github.com/open-webui/pipelines/blob/main/examples/filters/langfuse_filter_pipeline.py
     • Or use the local copy provided at: example/langfuse_filter_pipeline.py

2. Access Langfuse:

   • Open your browser and go to http://localhost:4000.

3. Create an Admin Account and Project:

   • Create an admin account and then create an organization and a project.
   • Go to Project Settings and create an API key.
   • Retrieve the secret key and public key.

Open-WebUI Setup

1. Access Open-WebUI:

   • Open your browser and go to http://localhost:3000.

2. Create an Admin Account:

   • Create an admin account.

3. Upload the Pipeline Script:

   • Go to Settings -> Admin Settings -> Pipelines.
   • If online, paste this URL:

     https://raw.githubusercontent.com/open-webui/pipelines/refs/heads/main/examples/filters/langfuse_filter_pipeline.py
     

     into the Install from Github URL field and click the download button.
   • If offline or using a custom script, upload langfuse_filter_pipeline.py from your local machine via the Upload Pipeline section.

4. Configure the Script:

   • After uploading the pipeline, edit its configuration in the UI.
   • Replace the placeholder values as follows:
     • your-secret-key-here → your Langfuse secret key
     • your-public-key-here → your Langfuse public key
     • https://cloud.langfuse.com → http://langfuse-web:4000 (local address)

5. Monitor Usage:

   • You can now monitor Open-WebUI usage statistics from Langfuse.

Model Downloading

1. Access Open-WebUI:

   • Open your browser and go to http://localhost:3000.

2. Create an Admin Account:

   • Create an admin account if you haven’t already.

3. Pull Models:

   • Navigate to Settings -> Admin Settings -> Models.
   • Enter a model tag to pull from the Ollama library (e.g., phi3:mini).
   • Press the pull button.