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

Code Generation on MBPP (Accuracy, Avg.)

63.49Accuracy

Forward KL

33.236441.090748.94556.7993May 25, 2026May 26, 2026May 27, 2026May 28, 2026
Updated 1d ago

Evaluation Results

MethodLinks
2026.05
63.4946.76
2026.05
63.1247.89
2026.05
62.1746.34
2026.05
60.8547.29
2026.05
60.3244.66
2026.05
60.0547.09
2026.05
43.6-
2026.05
43.6-
2026.05
43.6-
2026.05
43.2-
2026.05
43.2-
2026.05
43-
2026.05
42-
2026.05
41.8-
2026.05
41.2-
2026.05
40.8-
2026.05
40.6-
2026.05
40.4-
2026.05
40.4-
2026.05
38.1-
2026.05
38-
2026.05
38-
2026.05
37.2-
2026.05
36.6-
2026.05
36.2-
2026.05
34.8-
2026.05
34.4-
2026.05
-38.8
2026.05
-36.5
2026.05
-36.5
2026.05
-40.5
2026.05
-42.7
2026.05
-42
2026.05
-43.3
2026.06
-19.8
2026.06
-17.8
2026.06
-23.2
2026.06
-24.4
2026.06
-22.8
2026.06
-38.84
2026.06
-35.2
2026.06
-32.8
2026.06
-33.2
2026.06
-0
2026.06
-33
2026.06
-34
2026.06
-35