List your company

Safety Behavior Evaluation on Safety evaluation suite

16.5ASR+ Safety 40k
14.2429.49544.7560.005May 12, 2026

Evaluation results

MethodLinks
+ Safety 40kSFT training data category=Trained on general SFT data + safety SFT data2026.0516.54370.25
+ Safety 4k + Random synth 4kSFT training data category=Trained on general SFT data + safety SFT data + SAE synthetic data2026.05203672
+ Safety 120kSFT training data category=Trained on general SFT data + safety SFT data2026.052121.578.75
+ Safety 8kSFT training data category=Trained on general SFT data + safety SFT data2026.052234.571.75
+ Safety 200kSFT training data category=Trained on general SFT data + safety SFT data2026.05241978.5
+ Safety 4k + Feature synth 4kSFT training data category=Trained on general SFT data + safety SFT data + SAE synthetic data2026.052420.577.75
Alpaca 50kSFT training data category=Trained on general SFT data only2026.05733.561.75