All work (9)

Result: 14/15

exact on a 15-item extraction eval

Result: 211

unit tests, none of which need an API key

Result: 82

offline unit tests, with an exact McNemar test per intent

Result: v0.0.1

In development

Result: 4.94 to 2.90

validation loss over 2,500 CPU training steps

Result: 111

offline tests

Result: 0.78

mAP@0.5 on a YOLOv8n plate detector

Result: 0.547

ROC-AUC, best of 6 classifiers on 1,095 matches

Result: Negative result

published anyway