Overview
Learn how to implement observability for Large Language Models (LLMs) in this 17-minute conference talk from OSACon. Explore the fundamental concepts of LLM observability and discover how OpenTelemetry integrates with AI systems to enhance monitoring capabilities. Gain insights into the newly adopted LLM semantic convention by the OpenTelemetry community and understand its practical applications. Dive into OpenLLMetry's approach to implementing comprehensive observability across the entire AI stack, including vector databases, LLMs, and model orchestration platforms. Master the techniques for systematic data collection and analysis that enable improved LLM performance, bias detection and correction, troubleshooting, and system reliability assurance.
Syllabus
Observability for Large Language Models with OpenTelemetry
Taught by
OSACon