#evaluation-metrics

#evaluation-metrics

"When you hire someone for your team, do you only look at their test scores and the speed they work at? Of course not."

In few-shot learning settings, the evaluation metrics must reflect the data imbalance commonly observed, containing largely fewer samples per class leading to skewed results; thus, specialized metrics are vital.

In evaluating LLM-generated mutations, we designed metrics that encompass cost, usability, and behavior, recognizing that higher mutation scores don't guarantee higher quality.