Samanda Operational Diagram

Preprocessing

  • A unified development and operations platform for everything from data preprocessing to application services (either on-premises or in the public cloud) is provided.

Refinement Loop

  • Specialized chatbot services are generated by fine-tuning the Polyglot-Ko model based on data provided by the client.
  • Research of transfer learning and federated learning techniques to prevent exposure of customer data during fine-tuning is currently in progress.

Model Services

  • Plugins that integrate the Chatbot withthe Customer Service API are developed and provided.

Samanda is a Large Language Model Operations (LLMOps) artificial intelligence service using OpenAssistant that aims to build a conversational interface that provides a high level of convenience and interactivity.

OpenAssistant, on which Samanda is based, is a chat-based conversational AI built on principles such as using open source platforms, utilizing machine learning models and techniques, and leveraging NLP to process user input. These features enable OpenAssistant to meet the needs of a wide range of end users through continuous improvement through collaboration among contributors, and to achieve significantly better performance than other language generation platforms such as ChatGPT.

With the combination of OpenAssistant and Quantum C&S's cloud technology, Samanda will be an adaptable artificial intelligence that meets the needs of today's customers.