Farfalle

Preview：

Introduce：

Farfalle is an open source AI-powered search engine that allows users to run native Large language models (LLMs) or use cloud models. It is based on the Perplexity clone and can be integrated with a variety of technology stacks such as the Next.js front end, FastAPI back end, Tavily search API, etc. It also provides documentation and demo videos for customizing Settings to help users get started quickly.

Stakeholders:
Farfalle is for developers and technologists, especially those interested in AI-powered search technology. It provides an open source solution that enables individuals and enterprises to leverage large language models to enhance search capabilities, whether on-premises or in the cloud.
Usage Scenario Examples:

• Developers can use Farfalle to create personalized AI search engines.
• Businesses can deploy Farfalle to provide internal search services and improve employee productivity.
• Educational institutions can use Farfalle as a teaching tool to help students gain better access to information.

The features of the tool:

• Support for natively running large language models such as llama3, gemma, mistral.
• Support cloud models such as Groq/Llama3, OlienAI/glit4-o.
• Provides Docker deployment Settings for easy deployment.
• Integrated Tavily search API with no external dependencies.
• Use Logfire for logging.
• Use Redis for rate limiting.
• Provides detailed documentation of custom Settings.

Steps for Use:

• First, make sure you have Docker and Ollama installed (if running a native model).
• Download one of the supported models: llama3, mistral, or gemma.
• To start the Ollama server, run ollama serve.
• Get API keys for Tavily, OlienAI (optional), and Groq (optional).
• Clone Farfalle warehouse to local.
• Add the necessary environment variables to the.env file.
• Run containers using Docker Comliose.
• Access httli://localhost:3000 to view the application.

Tool’s Tabs: AI search engine,search-engine