Libby D. Bot

Libby the librarian. AI agent specialized in creating and querying embeddings for RAG (Retrieval Augmented Generation).

Installation

You can install Libby D. Bot using pip:

pip install -U libby

Usage

Libby provides several commands through its CLI interface:

Creating Embeddings

Create embeddings from your documents in a specified directory:

libby embed --corpus_path /path/to/your/documents --collection_name your_collection

The corpus_path defaults to the current directory if not specified. The collection_name parameter allows you to organize your embeddings into different collections (defaults to 'main').

Querying Documents

After creating embeddings, you can ask questions about your documents:

libby answer "What is the main topic of the documents?" --collection_name your_collection

Generating Content

You can use Libby to generate content based on prompts:

# Generate using direct prompt
libby generate "Write a summary of..." --output_file output.txt

# Generate using prompt from file
libby generate "" --prompt_file input_prompt.txt --output_file output.txt

Features

• Multiple language support (English and Portuguese)
• Various AI models available (Llama3, Gemma, ChatGPT)
• PDF document processing and embedding
• Question answering with context from your documents
• Content generation capabilities

Configuration

Libby supports different AI models and languages. You can configure these through environment variables or the config.yml file.

Available Models:

• Llama3 (default)
• Gemma
• Llama3-vision
• ChatGPT

Supported Languages:

• English (en_US)
• Portuguese (pt_BR)

License

This project is licensed under the GPLv3 License.