bug
fixed
Model AUC drops to 0.72 after retrain
bill@transcom.net (Transcom IS) ยท
2026-04-12 11:05
Retrained the LightGBM model on a fresh corpus extracted from eFa archives. Initial test had AUC 0.998 but after a config change it dropped massively. Suspect feature extraction mismatch.
Comments (1)
bill@transcom.net (Transcom IS)admin
2026-04-13 09:40
extract-features-v2.py wasn't mirroring the scanner's featurize() function exactly โ drift in one-hot encoding for content-type. Synced both, back to 0.998 AUC.
Sign in to add a comment.