Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
STEM Reasoning on OCW
Loading...
17.3
Accuracy
AGPO
7.836
10.293
12.75
15.207
May 20, 2026
Accuracy
Updated 13d ago
Evaluation Results
Method
Method
Links
Accuracy
AGPO
clipping=adaptive, ATS...
2026.05
17.3
GRPO
ATS=true
2026.05
16.3
AGPO
clipping=adaptive, ATS...
2026.05
16.2
GRPO
clipping=fixed ε
2026.05
15.4
Adaptive-KL PPO
2026.05
14.6
DPO
2026.05
9.1
PPO
2026.05
8.2
Feedback
Search any
task
Search any
task
