Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Mathematical Reasoning on Minerva-Math (Accuracy %)
Loading...
57.4
Accuracy
Nemotron-49B (Teacher)
16.8192
27.3546
37.89
48.4254
Aug 5, 2025
Sep 28, 2025
Nov 21, 2025
Jan 14, 2026
Mar 9, 2026
May 2, 2026
Jun 26, 2026
Accuracy
Updated 25d ago
Evaluation Results
Method
Method
Links
Accuracy
Nemotron-49B (Teacher)
Model Role=Teacher, Ca...
2026.06
57.4
KL Annealing only
SEAD Components=KL Ann...
2026.06
56.6
Token Zones + KL Annealing
SEAD Components=Token...
2026.06
54.8
Token Zones only
SEAD Components=Token...
2026.06
54
GRPO
Category=Baselines
2026.06
53.7
OPD (RKL)
Category=Baselines
2026.06
53.3
Nemotron-8B (Student)
Model Role=Student, Ca...
2026.06
52.9
OPSD
Category=Baselines
2026.06
49.6
VRPO
Base Model=Qwen3-8B
2025.08
31.99
λ-GRPO
Base Model=Qwen3-8B
2025.08
31.62
KTAE
Base Model=Qwen3-8B
2025.08
29.41
Base
Base Model=Qwen3-8B
2025.08
28.68
VRPO
Base Model=Qwen2.5-7B-...
2025.08
19.85
KTAE
Base Model=Qwen2.5-7B-...
2025.08
19.12
Base
Base Model=Qwen2.5-7B-...
2025.08
18.75
λ-GRPO
Base Model=Qwen2.5-7B-...
2025.08
18.38
Feedback
Search any
task
Search any
task