List your company

Safety Evaluation on XSTest (Accuracy and Intervention Rate)

96AccuracyGuidance Model
-3.8422.084873.92Apr 22, 2026

Evaluation results

MethodLinks
Guidance ModelModel Pair=G→L, Pair Family=Cross-Family2026.0496—
Guidance ModelModel Pair=G→G, Pair Family=Within-Family2026.0496—
Guidance ModelModel Pair=L→G, Pair Family=Cross-Family2026.0494—
Guidance ModelModel Pair=L→L, Pair Family=Within-Family2026.0494—
Guidance ModelModel Pair=Q→L, Pair Family=Cross-Family2026.0460—
Guidance ModelModel Pair=Q→G, Pair Family=Cross-Family2026.0460—
Aligned base modelModel Pair=Q→L, Pair Family=Cross-Family2026.0415—
Aligned base modelModel Pair=G→L, Pair Family=Cross-Family2026.0415—
BlendInModel Pair=L→G, Pair Family=Cross-Family2026.041520.3
Aligned base modelModel Pair=L→L, Pair Family=Within-Family2026.0415—
Aligned base modelModel Pair=Q→G, Pair Family=Cross-Family2026.0414—
Aligned base modelModel Pair=L→G, Pair Family=Cross-Family2026.0414—
Aligned base modelModel Pair=G→G, Pair Family=Within-Family2026.0414—
BlendInModel Pair=G→G, Pair Family=Within-Family2026.041428
NUDGINGModel Pair=L→L, Pair Family=Within-Family2026.041222.2
BlendInModel Pair=L→L, Pair Family=Within-Family2026.041222.2
NUDGINGModel Pair=L→G, Pair Family=Cross-Family2026.041020.3
NUDGINGModel Pair=G→G, Pair Family=Within-Family2026.041028
NUDGINGModel Pair=G→L, Pair Family=Cross-Family2026.04827.5
BlendInModel Pair=G→L, Pair Family=Cross-Family2026.04827.5
BlendInModel Pair=Q→L, Pair Family=Cross-Family2026.04433.1
NUDGINGModel Pair=Q→L, Pair Family=Cross-Family2026.04333.1
NUDGINGModel Pair=Q→G, Pair Family=Cross-Family2026.04234.3
BlendInModel Pair=Q→G, Pair Family=Cross-Family2026.04234.3
Base ModelModel Pair=Q→G, Pair Family=Cross-Family2026.041—
Base ModelModel Pair=L→G, Pair Family=Cross-Family2026.041—
Base ModelModel Pair=G→G, Pair Family=Within-Family2026.041—
Base ModelModel Pair=Q→L, Pair Family=Cross-Family2026.040—
Base ModelModel Pair=G→L, Pair Family=Cross-Family2026.040—
Base ModelModel Pair=L→L, Pair Family=Within-Family2026.040—