Typing Tiny Stories

How to Play

1. Begin the Story: Click anywhere on the screen to start the game.
2. Type the Text: A story will appear one line at a time. Type the text exactly as you see it.
3. Correct Mistakes: Any errors you make will be highlighted in red. Use the Backspace key to go back and fix them.
4. Advance the Line: Each line of the story ends with a space. Typing the final space will complete the line and present the next one.
5. Write Your Own Line: After you correctly type enough words, you will be prompted to write your own single-line addition to the story.
6. Watch the AI Continue: Once you submit your line, the LLM will generate the next part of the story, building directly upon your contribution for a unique, collaborative experience.

Changelog

Version 1.1

This update focuses on enhancing the user experience with significant visual, auditory, and technical improvements.

• Dynamic Effects: Added particle effects, glowing text, and cursor animations to create a more polished and engaging visual experience.
• Improved Feedback: Enhanced visual cues for typing, making it easier to track progress and errors.
• Expanded Stories: Increased the variety of starting story prompts and improved the quality of the AI-generated text.
• Smoother Typing: Refined the text input system, including improved backspace handling for a more fluid and intuitive typing experience.
• Performance Optimization: Restructured the game state management and refactored the text generation logic to be synchronous, resulting in better performance and reliability.
• Code Organization: Improved maintainability by moving all constants, resource paths, and configuration settings into a dedicated config.h file.

Version 1.0 (Initial Release)

• Core Functionality: Initial public release of the proof-of-concept application.
• Interactive Storytelling: Implemented the core gameplay loop where players type AI-generated stories and can contribute their own lines for the LLM to continue.
• LLM Integration: Integrated a llama2-based LLM trained on the TinyStories dataset, optimized for real-time generation on local hardware.
• Foundation: Built with raylib for graphics and designed for WebAssembly (WASM) deployment.