Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Preference Alignment on PRISM normalized-step (test)
Loading...
2.328
Borda Avg
Hard Panel
1.75704
1.90527
2.0535
2.20173
Feb 4, 2026
Borda Avg
Copeland Score
Updated 1mo ago
Evaluation Results
Method
Method
Links
Borda Avg
Copeland Score
Hard Panel
Backbone=Llama-3.2-1B,...
2026.02
2.328
9,836.294
US-Rep
Backbone=Llama-3.2-1B,...
2026.02
2.171
5,125.712
Soft Panel
Backbone=Llama-3.2-1B,...
2026.02
1.88
-3,600.814
Full PRISM
Backbone=Llama-3.2-1B,...
2026.02
1.842
-4,729.654
Base
Backbone=Llama-3.2-1B,...
2026.02
1.779
-6,631.538
Feedback
Search any
task
Search any
task