Skip to content

README.md

Latest commit

 

History

History
136 lines (101 loc) · 3.84 KB

README.md

File metadata and controls

136 lines (101 loc) · 3.84 KB

Book Writer System

A comprehensive book development system that helps authors create, organize, and refine book content through a structured workflow.

Features

  1. Outline Creation

    • Create hierarchical book outlines in YAML/JSON format
    • Structure books with parts, chapters, and subtopics
    • AI-powered outline generation using DeepSeek R1:8B model

  2. Note Processing Pipeline

    • Embed notes using BGE-M3 model
    • Store in ChromaDB with metadata for organization

  3. Content Expansion

    • Expand raw notes into polished text using Stable Beluga 13B model
    • Auto-classify content into appropriate sections using DeepSeek R1:8B model
    • Multiple writing styles supported (academic, narrative, technical, conversational)

  4. Retrieval-Augmented Writing (RAG)

    • Advanced RAG loop for context-aware writing using Ollama models
    • Maintains content coherence throughout the book
    • Uses Stable Beluga 13B model for content generation

  5. AI Function Calling & Tool Integration

    • Support for AI models to call specific tools and functions
    • Predefined tools for note searching, content classification, and outline validation
    • Extensible tool registry for adding custom functions

  6. Book Assembly

    • Merge chapters according to outline
    • Generate complete Markdown manuscript
    • Support continuous refinement

  7. Export Options

    • Export to PDF format
    • Export to ePub format

Model Configuration

The system uses Ollama to run local language models. By default, it uses:

  • Content Expansion & RAG: stable-beluga:13b - Optimized for generating rich, detailed content
  • Outline Generation & Organization: deepseek-r1:8b - Optimized for structure and organization

Prerequisites

  1. Install Ollama on your system
  2. Pull the required models:
ollama pull stable-beluga:13b
ollama pull deepseek-r1:8b

Configuration

The model configuration can be customized in book_writer/model_config.yaml. The default configuration is:

models:
  content_expansion:
    model_name: "stable-beluga:13b"
    api_type: "ollama"
    temperature: 0.7
    max_tokens: 1024
    top_p: 0.9
  outline_generation:
    model_name: "deepseek-r1:8b"
    api_type: "ollama"
    temperature: 0.5
    max_tokens: 512
    top_p: 0.8
  organization:
    model_name: "deepseek-r1:8b"
    api_type: "ollama"
    temperature: 0.4
    max_tokens: 512
    top_p: 0.8
ollama:
  base_url: "http://localhost:11434"

Installation

pip install -r requirements.txt

Usage

Interactive Mode

python main.py

Follow the interactive prompts to create and manage your book project.

Testing the Ollama Integration

To test if your Ollama models are properly configured:

python test_ollama_models.py

This will run comprehensive tests including:

  • Model connectivity
  • Content expansion functionality
  • Outline generation
  • Organization/classification
  • Full integration
  • RAG functionality
  • Function calling capabilities

Model-Specific Task Assignment

  • Stable Beluga 13B: Used for content generation tasks (expanding notes into polished text, RAG generation)
  • DeepSeek R1 8B: Used for organizational tasks (outline generation, content classification)

AI Tool Integration

The system includes a flexible tool registry that allows AI models to call specific functions:

Available Tools:

  • search_notes: Search for notes based on a query
  • classify_content: Classify content into appropriate chapter and subtopic
  • expand_note: Expand a note into detailed content
  • validate_outline: Validate if an outline is properly structured

Adding Custom Tools:

You can register custom tools with the ToolRegistry to extend the system's capabilities.

This division allows the system to leverage the strengths of each model for the most appropriate tasks.