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

Code Reasoning on OJBench

10.34Accuracy

DFT

Updated 28d ago

Evaluation Results

Method	Links
DFT 2025.10		10.34	10.34
SFT 2025.10		9.91	9.91
Random 2025.10		9.91	9.91
VCORE 2025.10		9.48	9.48
iw-SFT 2025.10		6.9	6.9
DFT 2025.10		6.03	6.03
DFT 2025.10		5.6	5.6
iw-SFT 2025.10		5.6	5.6
Original 2025.10		5.6	5.6
SFT 2025.10		4.74	4.74
VCORE 2025.10		4.74	4.74
Original 2025.10		4.31	4.31
Random 2025.10		4.31	4.31
VCORE 2025.10		3.88	3.88
Random 2025.10		3.02	3.02
Original 2025.10		3.02	3.02
iw-SFT 2025.10		2.59	2.59
VCORE 2025.10		2.58	2.58
SFT 2025.10		2.16	2.16
Original 2025.10		1.29	1.29
SFT 2025.10		0.86	0.86
iw-SFT 2025.10		0.86	0.86
Random 2025.10		0.86	0.86
DFT 2025.10		0	0