List your company

Safety Evaluation on SafetyBench zh

83.2Avg ScoreROSE
34.8447.39559.9572.505Feb 19, 2024

Evaluation results

MethodLinks
ROSEbackbone=Qwen-chat-14B, zero-shot=true2024.0283.286.690.993.376.889.188.263.3
Qwen-chat-14Bdecoding=Regular, zero-shot=true2024.0282.886.791.493.475.48988.961.3
ROSEbackbone=Internlm-chat-20B, zero-shot=true2024.0281.282.686.890.375.781.583.470.4
Internlm-chat-20Bdecoding=Regular, zero-shot=true2024.028080.985.789.175.178.281.371.3
ROSEbackbone=Qwen-chat-7B, zero-shot=true2024.0278.379.983.689.275.474.980.466
Qwen-chat-7Bdecoding=Regular, zero-shot=true2024.0277.880.183.789.575.673.881.162.3
ROSEbackbone=Internlm-chat-7B, zero-shot=true2024.027776.382.187.273.573.178.968.8
Internlm-chat-7Bdecoding=Regular, zero-shot=true2024.0273.372.57982.567.267.775.868.4
ROSEbackbone=Chinese-Alpaca-13B, zero-shot=true2024.0272.171.878.383.669.571.177.257.1
Chinese-Alpaca-13Bdecoding=Regular, zero-shot=true2024.0269.970.278.282.56767.77750.5
ROSEbackbone=Chinese-Alpaca-7B, zero-shot=true2024.0264.46165.773.462.254.768.464.5
ROSEbackbone=Alpaca-13B, zero-shot=true2024.0264.361.370.17161.753.467.363.2
Chinese-Alpaca-7Bdecoding=Regular, zero-shot=true2024.0263.96065.472.960.555.768.863.6
Alpaca-13Bdecoding=Regular, zero-shot=true2024.0263.161.470.871.364.951.668.252.5
ROSEbackbone=Vicuna-13B, zero-shot=true2024.0259.655.161.168.558.354.766.954.8
Vicuna-13Bdecoding=Regular, zero-shot=true2024.0257.853.16167.252.250.467.654.9
ROSEbackbone=Alpaca-7B, zero-shot=true2024.0257.653.254.961.463.751.259.858
Alpaca-7Bdecoding=Regular, zero-shot=true2024.02565254.959.862.850.959.254.5
ROSEbackbone=Vicuna-7B, zero-shot=true2024.0250.946.645.156.355.343.552.755.3
Vicuna-7Bdecoding=Regular, zero-shot=true2024.0249.644.944.554.554.642.64855.4
Randomzero-shot=true2024.0236.736.4262849.534.527.649.9