Friendli Suite와 생성AI 서비스의 미래 - 최적화 기술과 혁신적 플랫폼

Learn about the future of generative AI services and Friendli Suite in this 20-minute conference talk from SK AI SUMMIT 2024. Discover how Friendli Suite's specialized optimization technology for generative AI models achieves faster speeds while reducing GPU costs. Explore the platform's capabilities in customized model training and optimized inference serving, which can cut inference time and maximize processing performance to reduce GPU costs by up to 90%. Examine real customer cases demonstrating how FriendliAI addresses challenges in training and serving generative AI models, while learning about their flexible product lineup including Dedicated Endpoints, Containers, and Serverless Endpoints with open-source model APIs like Llama 3.1. Gain insights from CTO Kyungin Yu, inventor of Continuous Batching technology, as he shares core technological innovations, performance improvements, and discusses the present and future of generative AI, along with FriendliAI's role in leading this technology.