Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Code Generation on HumanEval (pass@2, pass@4, pass@8)

59.15Pass@2

DLE (top-p+top-k)

45.827649.286352.74556.2037Apr 22, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.04
59.1567.0776.83
2026.04
59.1568.2976.83
2026.04
59.1569.5178.66
2026.04
57.9365.2473.17
2026.04
55.4964.4675.61
2026.04
55.4965.2471.95
2026.04
51.8365.8576.83
2026.04
49.3964.0273.78
2026.04
46.3457.9370.12