This project explores the limits of AI in generating web novels using Large Language Models (LLMs) and multi-agent systems. It aims to develop an application capable of automatically creating network novels, targeting AI enthusiasts and researchers interested in creative AI applications. The primary benefit is to push the boundaries of AI-driven narrative generation.

How It Works

The project addresses LLM limitations in long-form creative writing, specifically planning, memory, and originality. It proposes a solution inspired by RecurrentGPT, using LLMs to compress long texts into concise "memories" for iterative, loop-based text generation. Multi-agent collaboration and optimized prompts are employed to enhance originality and leverage LLM capabilities, guided by prior knowledge of web novel creation processes.

Quick Start & Requirements

• Install dependencies: pip install -r requirements.txt
• Implement LLM interface: Requires implementing the chatLLM function in LLM.py.
• Run demo: python demo.py saves results to novel_record.md.
• Run Gradio app: python app.py for a visual experience.
• Prerequisites: Requires a functional LLM API.

Highlighted Details

• Explores AI's potential in literary creation, building on existing advancements.
• Addresses LLM weaknesses in planning, long-text memory, and originality.
• Utilizes a multi-agent approach for enhanced creative output.
• Inspired by RecurrentGPT's cyclical computation for arbitrary text length.

Maintenance & Community

No specific information on contributors, sponsorships, or community channels is provided in the README.

Licensing & Compatibility

The README does not specify a license. Compatibility for commercial or closed-source use is undetermined.

Limitations & Caveats

The author states that current LLMs are not yet capable of generating full-length web novels. Future development directions include longer context windows, improved reasoning, RLHF with human feedback, and structured plot frameworks.