Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Toxicity and Diversity Evaluation on SafeNLP
Loading...
0.028
Toxicity
Mistral-7b-Instruct + CP
0.01588
0.09769
0.1795
0.26131
Jan 16, 2024
Toxicity
Dist-1
Dist-2
Dist-3
Updated 4d ago
Evaluation Results
Method
Method
Links
Toxicity
Dist-1
Dist-2
Dist-3
Mistral-7b-Instruct + CP
Backbone=Mistral-7b-In...
2024.01
0.028
0.09
0.41
0.68
Mistral-7b + CP
Backbone=Mistral-7b, E...
2024.01
0.043
0.3
0.6
0.72
CHRT
Backbone=Mistral-7b, E...
2024.01
0.132
0.1
0.19
0.21
CHRT
Backbone=GPT-2, Evalua...
2024.01
0.257
0.44
0.7
0.71
Mistral-7b-Instruct
Evaluation Mode=white-box
2024.01
0.269
0.18
0.54
0.76
Mistral-7b
Evaluation Mode=white-box
2024.01
0.331
0.32
0.59
0.65
Feedback
Search any
task
Search any
task