Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Code Generation on HumanEval (Accuracy@4)
Loading...
70
Accuracy@4
PAPO
33.912
43.281
52.65
62.019
Mar 27, 2026
Accuracy@4
Updated 2mo ago
Evaluation Results
Method
Method
Links
Accuracy@4
PAPO
Model=Qwen2.5-14B
2026.03
70
ORM(GRPO)
Model=Qwen2.5-14B
2026.03
66
PAPO
Model=Qwen2.5-7B
2026.03
63.9
PAPO
Model=Qwen3-4B-Base
2026.03
63.9
ORM(GRPO)
Model=Qwen2.5-7B
2026.03
61.6
Base
Model=Qwen2.5-14B
2026.03
56.9
Base
Model=Qwen2.5-7B
2026.03
54.6
ORM(DAPO)
Model=Qwen3-4B-Base
2026.03
53.8
PAPO
Model=Qwen2.5-3B
2026.03
52.5
Base
Model=Qwen3-4B-Base
2026.03
49.1
ORM(GRPO)
Model=Qwen2.5-3B
2026.03
39.3
Base
Model=Qwen2.5-3B
2026.03
35.3
Feedback
Search any
task
Search any
task
