This repository provides several agent applications that work with production stack and LMCache endpoints. Each agent is designed to demonstrate different AI capabilities and deployment patterns, all integrated with LMCache's caching infrastructure for optimal performance.
An intelligent dual-agent academic paper analysis system that compares reviews from two different LLM endpoints side-by-side, leveraging LMCache for efficient endpoint management.
Key Features:
- 🌐 Modern Gradio web interface
- 🔄 Dual-agent comparison system
- 📄 PDF processing and analysis
- 📊 Real-time progress tracking
- 🔄 LangGraph workflow orchestration
- 🏠 Flexible LLM support (OpenAI API + local models)
- ⚡ LMCache integration for optimized endpoint usage
A Dify-powered research paper chatbot that demonstrates performance comparison between LMCache and vLLM inference engines. Upload research papers and ask questions with side-by-side responses from both engines.
Key Features:
- 📄 Document upload (PDF, images)
- 🤖 Dual LLM comparison (LMCache vs vLLM)
- 📊 Performance metrics and token usage statistics
- 🔄 Real-time comparison interface
- 🎯 Research-focused Q&A capabilities
- ⚡ Direct performance benchmarking
This repository is actively expanding with additional agentic applications. Each new agent will include:
- Complete source code and documentation
- Production-ready Helm charts
- Docker containerization
- CI/CD pipeline integration
- Comprehensive testing suite
- LMCache endpoint integration
- CI/CD Pipeline Guide: Complete guide for automated builds and deployments
- Helm Repository Guide: Complete guide for using and publishing Helm charts
- Paper Review Agent: Detailed documentation for the paper review application
- Dify Paper Chat Agent: Complete setup guide for the Dify-powered research paper chatbot