PROJECT FOUR - EVALUATION AND OBSERVABILITY

LEARNING GOALS By the end of this chapter you will be able to add tracing so that every model call and tool call leaves a timed record. You will be able to build a simple dashboard that summarizes success, cost, and speed. You will be able to build and run a fifteen-item evaluation set. You will be able to find a runaway loop in a trace and fix it with a guard.