Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Reward Prediction on NuminaMath (out-of-domain)
Loading...
42.22
Accuracy
MemReward
31.1336
34.0118
36.89
39.7682
Mar 13, 2026
Accuracy
Updated 26d ago
Evaluation Results
Method
Method
Links
Accuracy
MemReward
Backbone=Qwen2.5-3B-In...
2026.03
42.22
R1-Oracle
Backbone=Qwen2.5-3B-In...
2026.03
39.33
R1-p
Backbone=Qwen2.5-3B-In...
2026.03
36.44
MemReward
Backbone=Qwen2.5-1.5B-...
2026.03
34.67
R1-Oracle
Backbone=Qwen2.5-1.5B-...
2026.03
32
R1-p
Backbone=Qwen2.5-1.5B-...
2026.03
31.56
Feedback
Search any
task
Search any
task