FIELD NOTES TOPIC
qrdata tasks
1 note, newest first.
Grading the Grader: Lessons from Evaluating an Agentic Data Analysis System
The research evaluates automated graders for agentic data analysis systems using LAMBDA, a multi-agent system, on 153 numerical QRData tasks. The study introduces a three-layer human-AI grading cascade and an iterative…
