Note Training mix behind the T2.1 MoE detector (84,000 rows)

Note Training mix behind the mixA MoE detector (145,569 rows)

woog/open-text-detector-qwen36-35b-a3b-oct10

Token Classification • Updated about 7 hours ago • 1

Note The model on opentextdetector.com (Oct 10). Trained on the Oct 7 72,000-window mix.

Note Training mix of the Oct 10 model (72,000 windows, built Oct 7).

Note Held-out evaluation sets behind the published results (evaluation only, do not train on them)