World’s largest cloud provider and e-commerce company

Summary

Centific fine-tuned a large-scale RAG system using LLMs for an interactive and safe conversational experience, helping customers find answers to product questions across categories.

Challenge

To achieve the organization’s goals, the client’s chatbot development team needed to improve the safety of the LLM’s responses prior to making it available for general consumer use.

Solution

Using Centific’s SafeAI framework, our teams annotated data-run workstreams of reinforcement learning from human feedback (RLHF) processes, supervised fine-tuning and red teaming, and created (or transcreated) a variety of content—including model instructions, golden sets, and prompt and response pairs—across a variety of modalities (e.g., text, image, audio, and video).

Results

The client achieved high levels of both safety and performance without having to compromise on one for the other. This made for a more fulfilling and enjoyable online shopping experience.