Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Safety Classification on Chinese Response Datasets

85.11Beavertails Score

Qwen3Guard-0.6B-Strict

60.77467.09273.4179.728Jun 13, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
85.1173.3382.6780.37
2026.06
85.0873.2882.6980.35
2026.06
80.7965.7665.3470.63
2026.06
79.6869.4888.3279.16
2026.06
79.4271.6773.5374.87
2026.06
79.2666.4786.777.48
2026.06
79.1255.0143.0459.06
2026.06
78.4774.9196.4283.27
2026.06
77.2254.854659.36
2026.06
77.1871.4685.7178.12
2026.06
75.3762.5863.0767.01
2026.06
73.4965.2183.8474.18
2026.06
71.0451.3992.5571.66
2026.06
66.9255.2966.6262.94
2026.06
65.0363.6878.2869
2026.06
61.7143.7386.7764.07