Skip to content

Latest commit

 

History

History
173 lines (131 loc) · 5.23 KB

README.md

File metadata and controls

173 lines (131 loc) · 5.23 KB

Ollama Model Lab

A web-based interface for testing and comparing different Ollama models with customizable parameters and prompts.

Ollama Model Lab Interface

Overview

Ollama Model Lab provides an intuitive playground for exploring and comparing different Ollama models. Unlike typical chat interfaces or benchmark tools, this lab environment allows you to:

  • Test multiple models simultaneously with the same prompt
  • Compare detailed performance metrics and response characteristics
  • Customize model parameters and observe their impact
  • Generate comprehensive comparison reports
  • Save and manage frequently used prompts

Example comparison report output:

### phi4:14b vs hermes3:8b vs qwen2.5:7b
Performance comparison:
- phi4:14b: 21.05s total (7.13s load)
- hermes3:8b: 6.52s total (4.45s load)
- qwen2.5:7b: 9.31s total (4.19s load)

Each model's response includes:
✓ Token counts and timing
✓ Parameter settings used
✓ Model architecture details
✓ Full response text

See example-report.md for a complete sample output.

Prerequisites

  • Ollama installed and running
  • At least one Ollama model pulled
  • Python 3.x (if using the included server script)
  • A modern web browser

Quick Start

  1. Get the files:

    • Download the latest release ZIP file from the Releases page
    • Extract the ZIP file to your desired location

    Alternative for contributors: Clone the repository

    git clone https://github.com/designcomputer/ollama-model-lab.git
cd ollama-model-lab

  2. Start a local web server:

    • Windows users can double-click start.bat
    • Or use any method to serve the files locally: 
      # Python 3.x (default port 80)
python -m http.server 80

# Use a different port if 80 is in use
python -m http.server 8080

# Python 2.x
python -m SimpleHTTPServer 80

# Or use any other local server of your choice

  3. Ensure Ollama is running (default: http://127.0.0.1:11434)

  4. Open your browser and navigate to:

    http://localhost           # if using port 80
http://localhost:8080      # if using port 8080 (or your chosen port)

Features

Model Management

  • View all available Ollama models
  • Select multiple models for testing
  • Sort models by name or size
  • Clear test selection with one click

Parameter Configuration

  • Override default model parameters:
    • Temperature
    • Context Window
    • Max Tokens
    • Top K
    • Top P
    • Number of GPUs
    • Memory Mapping

Prompt Management

  • Save frequently used prompts
  • Import/Export prompt collections
  • Quick selection of saved prompts
  • Example prompts included

Response Analysis

  • Side-by-side response comparison
  • Detailed performance statistics
    • Total processing time
    • Load time
    • Token counts
    • Processing durations
  • Model-specific information display
  • Generate detailed comparison reports

Usage Tips

  1. Selecting Models:

    • Use the left panel to choose available models
    • Click the right arrow to add them to your test set
    • Click the left arrow to remove models from testing

  2. Configuring Parameters:

    • Select a model in the right panel
    • Click the gear icon to open parameter settings
    • Enable only the parameters you want to override

  3. Managing Prompts:

    • Save useful prompts with descriptive names
    • Use the dropdown to quickly load saved prompts
    • Import/Export prompts via Settings

  4. Generating Reports:

    • Run your test across all selected models
    • Click "Download Report" to save a detailed markdown report
    • Reports include all responses and statistics

Settings

Access the settings modal to:

  • Configure the Ollama server URL
  • Import/Export saved prompts
  • View version information

Troubleshooting

Port Configuration

  • If port 80 is already in use (common with web servers or other services):
    1. Modify start.bat to use a different port number
    2. Or start the server manually with a different port: python -m http.server 8080
    3. Remember to access the application using the correct port in your browser URL

Common Issues

  • Ensure Ollama is running before starting the application
  • Check the Ollama server URL in settings if models aren't loading
  • Clear your browser cache if you experience UI issues
  • Make sure your chosen port isn't blocked by firewall settings

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

  • Built for use with Ollama
  • Uses browser IndexedDB for prompt storage
  • Inspired by the need for easy model comparison

Support

If you encounter any issues or have questions:

  1. Check the Issues page
  2. Submit a new issue with detailed information about your problem
  3. Include your browser and Ollama versions when reporting issues