Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Safety Evaluation on Safety Benchmarks Aggregate (test)
Loading...
73.6
Generation Quality (Std Prefix)
Self-Improving Pretraining
48.016
54.658
61.3
67.942
Jan 29, 2026
Generation Quality (Std Prefix)
Generation Quality (Unsafe Prefix)
Standard Evals (Avg)
Safety Evals (Avg)
Updated 4d ago
Evaluation Results
Method
Method
Links
Generation Quality (Std Prefix)
Generation Quality (Unsafe Prefix)
Standard Evals (Avg)
Safety Evals (Avg)
Self-Improving Pretraining
Backbone=Llama 1.4B, T...
2026.01
73.6
77.7
49.1
91.1
Llama Pretrain Baseline
Backbone=Llama 1.4B, T...
2026.01
54.5
52.6
47.9
75.5
Llama Base
Backbone=Llama 1.4B, P...
2026.01
50
50
47.6
76.9
Llama Pretrain Baseline
Backbone=Llama 1.4B, T...
2026.01
49
44.9
46.8
77
Feedback
Search any
task
Search any
task