Annotation Entropy Predicts Per-Example Learning Dynamics in LoRA Fine-Tuning

Build a diagnostic dashboard for LoRA fine-tuning that detects 'un-learning' on controversial training data. This helps devs prune datasets that induce performance degradation.

Suggested repo: EntropyTune

"Find out which training examples are actually ruining your fine-tuning run."

Estimated effort: 25h