Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Prosocial Alignment on HarmEval (test)
Loading...
76.3
MIP
PROSOCIALALIGN
42.812
51.506
60.2
68.894
Dec 6, 2025
MIP
Updated 4d ago
Evaluation Results
Method
Method
Links
MIP
PROSOCIALALIGN
Backbone=llama
2025.12
76.3
PROATTR-GEN-PCA
Backbone=mistral
2025.12
70.9
DIREG
Backbone=llama
2025.12
67.3
PP
Backbone=llama
2025.12
67
PROATTR-GEN-PCA
Backbone=llama
2025.12
64.8
PROSOCIALALIGN
Backbone=mistral
2025.12
64.3
CTRL-GEN
Backbone=llama
2025.12
62.5
CTRL-GEN
Backbone=mistral
2025.12
60.7
DIREG
Backbone=mistral
2025.12
59.8
PV-ARM-SUM
Backbone=llama
2025.12
57.6
PP
Backbone=mistral
2025.12
56.7
SAFE-ARITH
Backbone=llama
2025.12
52.5
SAFE-ARITH
Backbone=mistral
2025.12
45.1
PV-ARM-SUM
Backbone=mistral
2025.12
44.1
Feedback
Search any
task
Search any
task