RAG overview

Available in VPC

NAVER Cloud Platform's Retrieval-Augmented Generation (RAG) service leverages NAVER Cloud's hyperscale AI technology, the HyperCLOVA X language model, along with the highly durable and available Object Storage service, to help users build services with accurate and efficient search capabilities. With the RAG service, you can test customized search using your own data without training a separate large language model, and evaluate the results to verify performance.

RAG features

NAVER Cloud's RAG service offers these key features:

• Robust data security: You can safely use the service by configuring your search targets with your own validated data through API integration or data stored in Object Storage buckets.
• User-friendly RAG system: RAG reduces hallucination risks by combining RAG systems with LLMs. It provides a web console interface for easily testing data retrieval and evaluating datasets.
• Flexible model testing: You can run tests on search results before implementing your tailored search service. Various options are available for testing search models, allowing you to customize the text splitting methods, model configuration, and number of sources to be displayed for the retrieval process. In addition to search data, you can also customize LLM model integration, prompts, and parameters for generating answers.
• Source attribution policy: By indexing your configured data, RAG reduces hallucination risks and enables testing of tailored search services based on specific information needs.
• Data quality validation: You can easily run evaluations using your own test sets for a given test model and download the results.

RAG user guide

RAG is available in the Korea Region. Use this guide to get the most out of RAG.

• RAG overview: Learn about the service and find helpful resources.
• RAG prerequisites: View supported environments and pricing information.
• RAG quickstart: Follow a step-by-step guide through the entire process.
• Getting started: Learn how to get started with RAG from the NAVER Cloud Platform console.
  • Manage subscriptions: Subscribe to and manage your RAG.
• Using RAG:: Learn how to access RAG from the NAVER Cloud Platform console.
  • RAG interface: Discover the home interface of RAG in detail.
  • Services: Find how to create an RAG service from the NAVER Cloud Platform console.
    • Create service: Create, delete, and manage a service.
    • API key management: Create and manage API Keys.
    • Service testing: Learn how to run model tests for the created service.
  • Evaluations: Discover how to evaluate an RAG service.
    • Create evaluation: Create and delete RAG service model evaluations.
    • Run evaluation: Find how to run created evaluations.
• RAG resource management: Learn about RAG resources.
• RAG permissions management: Manage accounts using NAVER Cloud Platform’s Sub Account.
• RAG service AI ethics guide: Learn about the policies and practice plans applied to RAG for observing NAVER AI ethical principles.
• RAG content operation policy: Discover the policies on contents produced with RAG.
• RAG glossary: Find key terms and definitions.
• RAG release notes: See documentation updates.

RAG related resources

Beyond the user guide, these resources provide additional context and support for RAG. If you're considering adopting RAG or need in-depth information on the service, these resources can help:

• Learn about RAG development

  • RAG service API guide: A guide for developers on how to use APIs.

• Learn more about RAG

  • Ncloud user environment guide: A guide on the VPC environment and supported features.

• Learn about integrated services

  • Sub Account user guide: A guide to using Sub Account when you need an administrator account with multiple permission levels to manage the RAG service.
  • Object Storage user guide: A guide to using Object Storage as the data storage to be indexed when creating an RAG service.
  • CLOVA Studio user guide: A guide to using CLOVA Studio, which provides the LLM models required for answer generation in RAG services.