Skip to content

Releases: TrentPierce/PolyCouncil

PolyCouncil v1.1.0

20 Jan 15:06

Choose a tag to compare

PolyCouncil v1.1.0 Release Notes

Release Date: January 20, 2026


What's New

Stability & Reliability Improvements

  • Thread-Safe Settings: Fixed a race condition that could cause settings to be lost when toggling multiple options quickly
  • Single-Voter Validation: The app now verifies that the selected single-voter model is actually loaded before using it, with a helpful warning if not
  • Better Error Messages: Clearer feedback when issues occur, especially around model compatibility and voting

Fairness Improvements

  • Fair Tie-Breaking: When multiple models score equally, the winner is now randomly selected instead of always picking the first one. This eliminates positional bias that gave an unfair advantage to models selected earlier.
  • Smarter Consensus Detection: Discussion mode now actually analyzes message content for agreement signals (words like "agree", "consensus", "conclude") rather than just counting turns

User Experience

  • Keyboard Shortcuts:

    • Ctrl+Enter - Send message to council
    • Ctrl+Shift+A - Select all models
    • Ctrl+R - Refresh models list
    • Escape - Stop current operation
  • Cleaner UI: The concurrency warning only appears when you set concurrency above 2 (no more unnecessary clutter)

  • Enhanced Leaderboard: Now shows win percentages alongside win counts (e.g., "ModelName — 5 wins (42%)")

Bug Fixes

  • Fixed stat_view.py which had a hardcoded path that only worked on the developer's machine
  • Fixed PySide6 compatibility issue with keyboard shortcuts

Installation

Standalone Executable (Recommended)

  1. Download PolyCouncil.exe from this release
  2. Run the exe (no installation required)
  3. Make sure LM Studio is running with at least one model loaded
  4. Click "Connect" and start deliberating!

From Source

git clone https://github.com/TrentPierce/PolyCouncil.git
cd PolyCouncil
pip install -r requirements.txt
python council.py

Acknowledgments

Thanks to everyone who provided feedback and reported issues. This release addresses the findings from a comprehensive code audit focused on stability, fairness, and user experience.


Full Changelog: v1.0.1...v1.1.0

Full Changelog: v1.0.1...v1.1.0

PolyCouncil v1.0.1 - Icon Fix Release

17 Nov 14:57
886e999

Choose a tag to compare

PolyCouncil v1.0.1 - Icon Fix Release

What's New

Icon Fixes

  • Fixed Icon Color: Resolved color accuracy issues in the application icon
  • Fixed Icon Scaling: Improved icon scaling across different Windows display contexts and resolutions
  • Better Icon Quality: Updated to use optimized icon file for better visual fidelity

Changes

  • Updated EXE build configuration to use PolyCouncilIco.ico for improved color accuracy and proper scaling
  • Icon now displays correctly with the intended blue color scheme
  • Multi-resolution icon support ensures proper display at all sizes

Installation

  1. Download PolyCouncil.exe from this release
  2. Replace your existing PolyCouncil.exe with the new version
  3. Run the application - the icon should now display correctly

Note: If the icon still shows incorrectly, Windows may have cached the old icon. Try:

  • Restarting File Explorer
  • Moving the EXE to a different location
  • Clearing Windows icon cache

Technical Details

This release addresses icon rendering issues where the programmatically generated icon had color channel conversion problems and scaling artifacts. The fix uses a pre-optimized icon file that ensures proper color representation and scaling across all Windows display contexts.

Compatibility

  • Same system requirements as v1.0.0
  • Windows 10/11 (64-bit)
  • LM Studio with at least one model loaded

Enjoy the improved icon display!

PolyCouncil v1.0.0 - First Release

12 Nov 18:28

Choose a tag to compare

PolyCouncil v1.0.0 - First Release 🎉

🚀 What is PolyCouncil?

PolyCouncil is a multi-model deliberation engine for LM Studio that runs multiple local LLMs in parallel, gathers their responses, and has them evaluate each other using a shared scoring rubric. After scoring, the models vote on which answer is best, producing a consensus-driven final result.

Running a single model gives you one perspective. Running a council gives you… perspective squared.

✨ Key Features

🎯 Core Functionality

  • Parallel Model Execution: Run several local models simultaneously and collect answers in parallel
  • Rubric-Based Scoring: Each model scores every other response across customizable criteria
  • Consensus Voting System: Final answer is chosen through a weighted vote based on rubric scores
  • Leaderboard Tracking: Track which models perform best over time

🎭 Persona System

  • Custom Personas: Create and assign different personas to different models
  • Built-in Personas: Includes recommended personas like "Meticulous fact-checker", "Pragmatic engineer", "Cautious risk assessor", and more
  • Per-Model Assignment: Apply different personas to different models simultaneously to test varying perspectives

⚖️ Judge Mode

  • Single-Voter Mode: Optionally use a single model to act as the "ultimate judge" for all votes
  • Flexible Voting: Choose which models participate in voting vs. answering

🔧 Advanced Features

  • Concurrent Processing: Adjustable concurrency for faster processing (1-8 concurrent jobs)
  • Debug Tools: Toggle verbose logging for internal scoring/behavior analysis
  • Settings Management: Centralized settings dialog for easy configuration
  • Model Discovery: Automatic detection of available models from LM Studio

📦 What's New in v1.0.0

🎁 Standalone Windows Executable

  • No Python Required: Download and run PolyCouncil.exe directly - no installation needed!
  • All Dependencies Included: Everything bundled in a single executable (~36 MB)
  • Custom Icon: Beautiful blue icon matching the application theme

🎨 UI Improvements

  • Settings Dialog: Centralized settings with persona management
  • Persona Buttons: Easy-to-use persona selection buttons for each model
  • Debug Log Dock: Dockable debug log panel for detailed analysis
  • Modern Interface: Clean, intuitive GUI built with PySide6

🐛 Bug Fixes & Improvements

  • Fixed persona system visibility and functionality
  • Improved model list refresh stability
  • Enhanced icon display on Windows
  • Better error handling and user feedback

📋 System Requirements

  • Operating System: Windows 10/11 (64-bit)
  • LM Studio: Must be installed and running with at least one model loaded
  • RAM: Recommended 8GB+ (depends on your models)
  • Storage: ~50 MB for the executable

Note: LM Studio must be running and accessible at http://localhost:1234 (default) or your configured base URL.

🚀 Quick Start

  1. Download: Download PolyCouncil.exe from this release
  2. Run LM Studio: Start LM Studio and load at least one model
  3. Launch: Double-click PolyCouncil.exe
  4. Connect: Click "Connect" to discover available models
  5. Select Models: Check the models you want to use
  6. Ask Questions: Type your question and click "Send"

📖 Usage Tips

First Time Setup

  1. Open Settings (top-right button) to configure:
    • Enable personas if you want to use different perspectives
    • Adjust max concurrency based on your hardware
    • Enable debug logs for detailed analysis

Using Personas

  1. Go to Settings → Enable personas
  2. Click the "Persona" button next to each model
  3. Select a persona from the menu
  4. Different personas will give different perspectives on the same question

Single-Voter Mode

  • Check "Single-voter" in the main window
  • Select which model should act as the judge
  • All models will answer, but only the selected model will vote

Debug Mode

  • Enable in Settings → "Enable debug logs"
  • A dockable log panel will show detailed scoring and voting information
  • Useful for understanding how the council reaches consensus

🎯 Use Cases

  • Model Benchmarking: Compare how different models perform on the same questions
  • Research: Explore emergent behavior from model voting and deliberation
  • Reliability: Get more reliable answers by combining multiple model perspectives
  • Experimentation: Test different personas and voting strategies
  • Education: Understand how ensemble methods work in practice

⚠️ Known Issues & Notes

  • First Run: Windows Defender may scan the executable on first launch (this is normal)
  • Icon Cache: If the icon doesn't display correctly, Windows may need to refresh its icon cache (restart File Explorer)
  • LM Studio Connection: Ensure LM Studio is running before connecting
  • Model Loading: Large models may take time to respond - be patient!

🔗 Alternative Installation

Prefer to run from source? Clone the repository and install dependencies:

git clone https://github.com/TrentPierce/PolyCouncil
cd PolyCouncil
pip install -r requirements.txt
python council.py

📄 License

This project is licensed under the Polyform Noncommercial License 1.0.0.

  • ✅ Free for personal and research use
  • ❌ Commercial use is not permitted

🙏 Acknowledgments

Built for model tinkerers, curious minds, and anyone who enjoys watching artificial brains argue politely.

🐛 Report Issues

Found a bug or have a feature request? Please open an issue on GitHub:
https://github.com/TrentPierce/PolyCouncil/issues


Enjoy exploring the power of multi-model deliberation! 🎉