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

Reasoning on CALI OOD (test)

77.95Accuracy

MultiRole-R1

Updated 1mo ago

Evaluation Results

Method	Links
MultiRole-R1 2025.07		77.95	83.84
SelfConsis SFT+GRPO 2025.07		77.83	80.36
MultiRole-R1 2025.07		76.5	89.08
SelfConsis SFT+DPO 2025.07		76.1	69.58
SelfConsis SFT 2025.07		76.08	73.65
SelfConsis SFT 2025.07		75.96	72.78
More think 2025.07		75.9	76.81
SelfConsis SFT+DPO 2025.07		75.82	71.71
SelfConsis SFT+GRPO 2025.07		75.65	84.92
More think 2025.07		75.1	68.44
Zero-shot CoT 2025.07		75.05	71.83
Zero-shot CoT 2025.07		73.4	57.43
MultiRole-R1 2025.07		71.48	90.55
Self-Refine 2025.07		71.28	78.22
MultiRole-R1 2025.07		70.85	83.31
SelfConsis SFT+GRPO 2025.07		70.83	82.15
Role-Play 2025.07		70.26	50.49
SelfConsis SFT 2025.07		70.05	79.77
SelfConsis SFT+DPO 2025.07		69.81	76.62
Self-Refine 2025.07		69.4	48.16
SelfConsis SFT+GRPO 2025.07		69.26	83.37
SelfConsis SFT+DPO 2025.07		68.19	64.09
Role-Play 2025.07		67.41	70.74
SelfConsis SFT 2025.07		67.35	78.94
More think 2025.07		64.41	80.3
Role-Play 2025.07		62.7	77.32
Self-Refine 2025.07		61.95	78.87
Zero-shot CoT 2025.07		60.84	73.98
More think 2025.07		60.45	78.82
Role-Play 2025.07		52.69	67.43
Self-Refine 2025.07		50.76	66.09
Zero-shot CoT 2025.07		50.3	52.22