Building and Evaluating Prompts on Production Grade Datasets for Conversational AI

Building and Evaluating Prompts on Production Grade Datasets for Conversational AI

Learn how to effectively construct and evaluate prompts for production-level Large Language Model (LLM) implementations in this 29-minute conference talk from the Toronto Machine Learning Series. Explore methodologies and techniques for creating production-style datasets specifically designed for LLM tasks, with a focus on conversational AI applications. Discover practical insights from Voiceflow's Lead of Agent Performance & ML Platform Bhuvana Adur Kannan and Machine Learning Engineer Yoyo Yang as they share their experiences in developing and deploying prompt-based features. Master the challenges of prompt engineering in production environments while gaining valuable lessons learned from real-world implementations.