This repository provides a comprehensive collection of sample applications for Vespa.ai, a big data serving engine. It targets developers and researchers looking to leverage Vespa for various use cases, including vector search, hybrid search, RAG, and traditional text search, offering practical examples for both self-hosted and cloud deployments.

How It Works

The sample applications demonstrate Vespa's capabilities by showcasing its schema definition language, data feeding mechanisms, and query language. They highlight Vespa's support for advanced features like multi-vector indexing, hybrid search (combining keyword and vector search), and integration with various embedding models (e.g., Cohere, BGE-M3, OpenAI Matryoshka). The examples often utilize Python (Pyvespa) for interaction and data processing, emphasizing efficient data handling and scalable search.

Quick Start & Requirements

• Deployment: Samples can be deployed to Vespa Cloud or self-hosted. Specific deployment instructions are detailed within each sample's README.md.
• Dependencies: May include Python, specific libraries (e.g., pyvespa), and potentially GPU/CUDA for advanced AI models. Refer to individual sample READMEs for precise requirements.
• Resources: Setup time and resource needs vary significantly based on the complexity and data size of the sample.
• Documentation: Vespa.ai Documentation

Highlighted Details

• Extensive examples for vector search, including multi-vector indexing and various embedding models.
• Demonstrations of Retrieval Augmented Generation (RAG) with both external and local LLM integration.
• Tutorials on traditional text search, hybrid search, and advanced ranking techniques (e.g., ColBERT, cross-encoders).
• Showcases for e-commerce search, recommendations, and real-time features like search-as-you-type.
• Examples for using Vespa as an ML inference server with support for model versioning and custom code.

Maintenance & Community

• Maintained by the Vespa.ai team.
• Community resources and support channels are available via the main Vespa.ai website.

Licensing & Compatibility

• Samples are typically provided under a permissive license, but users should verify the license of each individual sample application.
• Compatible with Vespa.ai, which is designed for integration into various data pipelines and applications.

Limitations & Caveats

Some sample applications may not yet fully adhere to the described standard structure, requiring users to consult individual READMEs for specific instructions and potential variations.