SigLIP2 Tagger

Status

Threshold

0.01 0.99

General / Meta Threshold

0.01 0.99

Character / Copyright Threshold

0.01 0.99

Show Jeffreys posterior probabilities in results. Tag selection is always based on raw sigmoid + best_thr.

Display calibrated probabilities

min best_thr

Clamp best_thr to this floor. Tags with very low best_thr (untrained / noise-level) are raised to this value. Active only in Best-thr mode.

0.1 0.5

min best_F1

Exclude tags whose best_F1 is below this threshold (effectively untrained tags). Active only in Best-thr mode.

0 0.3

Raise Character/Copyright thresholds when the image is out-of-distribution (Mahalanobis distance > p95). Requires _ood_ref.npz alongside the model.

OOD detection

Apply per-tag calibration. Only active in Fixed threshold mode.

Calibration (legacy fixed-threshold mode)

ε (Jeffreys)

0.1 2

Prior strength (Beta-BB)

0.5 50