LoLLMs v18.0 (beta) Changelog

Now the messages can no longer render raw HTML. This allows a new protection against some code injection attacks. Since now lollms can run on docker.

LoLLMs v18.0 (alpha) Changelog

LoLLMs v18 'Matrix'

01010100 01001000 01000101 01010010 01000101 00100000 01001001 01010011 00100000 01001110 01001111 00100000 01010011 01010000 01001111 01001111 01001110

🎯 Major Features

Implemented new "Think First" process for all AI interactions
Added dedicated thinking visualization UI component
Integrated collapsible thinking section in chat interface
New: Automatic changelog popup notification system

🖥️ UI Improvements

New collapsible "Thinking Process" panel showing AI's reasoning steps
Added toggle button to show/hide thinking process
Enhanced step-by-step visualization with improved formatting
Implemented progress indicators for thinking stages
New styling for thought process separation
New: Changelog popup dialog
- Appears automatically when new version is detected
- Clean, modern design with scrollable content
- Dismissible with "Don't show again" option
- Accessible from settings menu for later viewing

🧠 AI Enhancements

Integrated systematic thinking approach before response generation
Added structured thinking patterns (hypothesis, analysis, planning)
Improved problem-solving capabilities through explicit reasoning steps
Enhanced decision-making transparency

⚡ Performance Optimizations

Optimized thinking process display rendering
Improved response generation pipeline
Better memory management for thinking process tracking

🛠️ Technical Updates

Added new API endpoints for thinking process monitoring
Implemented new thinking process data structure
Enhanced logging system for thought tracking
Added new configuration options for thinking visualization
New: Changelog version tracking system
- Local storage for changelog display preferences
- Version comparison utilities
- Automated popup triggering system

📝 Documentation

Added documentation for new thinking process features
Updated user guide with thinking visualization instructions
New examples showcasing think-first approach
New: Documentation for changelog notification system

Note

This update demonstrates that sophisticated thinking capabilities can be achieved with traditional LLMs through proper prompting and visualization, without requiring O1 or R1 models.

2.5 KiB Raw Blame History