You can deploy this model and start earning money today!

markprix12/
llama-rag-chatbot

Name: llama-rag-chatbot
Availability: InStock

Views

No views yet

Document Chat with Google Drive

Model Description

This application is a document-based question-answering system that integrates with Google Drive for document ingestion and uses LangChain with local LLMs for generating responses. It provides a Streamlit-based chat interface with source verification and relevancy scoring capabilities.

Model Architecture

The system uses a multi-component architecture:

Document Processing: Handles multiple file formats (PDF, DOCX, TXT, CSV, XLSX)
Embedding Model: sentence-transformers/all-MiniLM-L6-v2 for document embedding
Vector Database: FAISS for efficient document retrieval
Language Model: Local LLM (Ollama) with options for llama3 or mistral
Frontend: Streamlit-based interactive chat interface

Intended Use

Primary Use Cases

Document-based question answering
Enterprise knowledge base interaction
Customer service automation
Internal documentation search and retrieval

Out-of-Scope Uses

General conversation without document context
Processing of non-text-based media
Real-time document generation
Autonomous decision making

Performance and Limitations

Performance Metrics

Uses cosine similarity for relevancy scoring
Configurable chunk sizes (default: 512 tokens)
Chunk overlap: 256 tokens
Response length: 2-5 sentences (max 100 words)

Limitations

Requires Google Drive API credentials
Limited to text-based document formats
Requires local installation of Ollama
Memory constraints based on document volume

Training Data

This system doesn't require training as it uses:

Pre-trained sentence transformers for embeddings
Pre-trained local LLMs for response generation
Real-time document processing from user-provided sources

Technical Specifications

Model Dependencies

python

langchain_community>=0.0.10
langchain_core>=0.1.10
sentence_transformers>=2.2.2
faiss-cpu>=1.7.4
streamlit>=1.29.0
google-auth-oauthlib>=1.2.0
google-auth>=2.23.0
google-api-python-client>=2.100.0
pandas>=2.1.0

Hardware Requirements

Minimum 8GB RAM
Storage space dependent on document volume
CPU with AVX2 support recommended
GPU optional but recommended for larger document sets

Ethical Considerations

Privacy: All documents are processed locally
Data Security: Requires secure handling of Google Drive credentials
Transparency: Provides source documents for verification
Accuracy: Includes relevancy scoring for response validation

Deployment

Requirements

Google Drive API credentials (credentials.json)
Local Ollama installation
Python 3.8+
Required Python packages

Installation Steps

bash

# Clone the repository
git clone [repository-url]

# Install dependencies
pip install -r requirements.txt

# Set up Google Drive credentials
# Place credentials.json in the root directory

# Run the application
streamlit run app.py

Additional Information

Version Control

Version: 1.0.0
Last Updated: 2024-11-12
License: MIT

Contact Information

For support or questions, please open an issue in the GitHub repository.

Citation

If you use this code in your research, please cite:

bibtex

@software{document_chat_drive,
  title = {Document Chat with Google Drive},
  year = {2024},
  url = {[repository-url]},
  version = {1.0.0}
}

markprix12/llama-rag-chatbot