What is Marqo?
Marqo provides a comprehensive solution for training, deploying, and managing embedding models, offering blazingly fast image ingestion, inference, and search capabilities. The platform supports hundreds of embedding models out of the box and allows for custom weights implementation.
With features like horizontally scalable vector and lexical retrieval, metadata handling, reranking, and hybrid search, Marqo delivers advanced semantic search capabilities. The platform excels in multimodal search, supporting both text and image-based queries, while offering multilingual support across 100+ languages.
Features
- End-to-End Integration: Complete coverage from training to inference to storage
- Multimodal Search: Combined text and image search capabilities
- Multilingual Support: Search functionality in over 100 languages
- Scalable Architecture: Supports multi-terabyte indexes with low latency
- Model Training: Fine-tuning capabilities with Generalized Contrastive Learning
- Evaluation Tools: Comprehensive retrieval performance assessment
- Hybrid Search: Combined vector and lexical retrieval capabilities
Use Cases
- Advanced semantic search implementation
- RAG and generative AI applications
- Product recommendations and personalization
- Content moderation and duplicate detection
- E-commerce catalog optimization
- Multilingual search systems
- Image-based search applications
FAQs
-
What is vector search?
Vector search converts documents, images and other data into vectors that summarize the data in semantic form, allowing for matching documents against queries through analysis of semantic content and understanding where and how the document matched the query. -
Do I have to change my code to move from open-source to cloud?
The only changes needed are updating the URL and API key when accessing Marqo.
