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

Binary Safety Classification on UnsafeBench

35.5Sexual

Qwen2.5-7B

33.3647.80562.2576.695Mar 1, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.03
35.524.569.155.347.237.533.923.32341.7
2026.03
41.425.35757.972.252.154.966.76.552
2026.03
44.6013.223.53411.56.825022.7
2026.03
51.226.35059.674.225.22312.23.743.6
2026.03
62.4---------
2026.03
72.924.157.51543.953.245.261.348.447.3
2026.03
73.8---------
2026.03
75.729.242.6-74.9--60--
2026.03
77.72761.533.369.768.750.6664955.9
2026.03
78.5---------
2026.03
81.924.959.227.966.767.748.172.553.556.5
2026.03
8733.86745.474.872.961.576.553.167
2026.03
8950.670.555.2796266.774.963.372.2