List your company

Mathematical Reasoning on ASDiv Out of Distribution (Top-1 Accuracy)

89.1Top-1 Accuracy (maj@1)CausalRM
19.451237.533155.61573.6969Oct 24, 2024Jan 9, 2025Mar 27, 2025Jun 12, 2025Aug 28, 2025Nov 13, 2025Jan 29, 2026

Evaluation results

MethodLinks
CausalRMEvaluation Type=Downstream RLHF2026.0189.1
GoalRMEvaluation Type=Downstream RLHF2026.0188.4
Standard RMEvaluation Type=Downstream RLHF2026.0188
SFTEvaluation Type=Downstream RLHF2026.0182.4
SIKeD (All)Model=Gemma 7B2024.1070.76
SIKeD (Adaptive)Model=Gemma 7B2024.1070.59
CombinedModel=Gemma 7B2024.1069.11
SIKeD (Adaptive)Model=Qwen 1.5B2024.1068.98
CoTModel=Gemma 7B2024.1068.76
CoTModel=Qwen 1.5B2024.1068.76
SIKeD (All)Model=Qwen 1.5B2024.1068.55
PoTModel=Gemma 7B2024.1067.85
CombinedModel=Qwen 1.5B2024.1067.64
PoTModel=Qwen 1.5B2024.1066.94
L2MModel=Gemma 7B2024.1064.69
L2MModel=Qwen 1.5B2024.1063.69
SIKeD (Adaptive)Model=Gemma 2B2024.1059.05
SIKeD (All)Model=Qwen 0.5B2024.1058.61
SIKeD (Adaptive)Model=Qwen 0.5B2024.1058.44
PoTModel=Gemma 2B2024.1058.13
SIKeD (All)Model=Gemma 2B2024.1058
CombinedModel=Gemma 2B2024.1057.96
CombinedModel=Qwen 0.5B2024.1057.79
PoTModel=Qwen 0.5B2024.1056.83
CoTModel=Qwen 0.5B2024.1054.66
CoTModel=Gemma 2B2024.1054.01
L2MModel=Qwen 0.5B2024.1049.76
SIKeD (All)Model=SmolLM 1.7B2024.1048.16
SIKeD (Adaptive)Model=SmolLM 1.7B2024.1047.2
CombinedModel=SmolLM 1.7B2024.1046.77
InfoRMEvaluation Type=Downstream RLHF2026.0146.3
PoTModel=SmolLM 1.7B2024.1043.77
L2MModel=Gemma 2B2024.1043.47
CoTModel=SmolLM 1.7B2024.1030.37
L2MModel=SmolLM 1.7B2024.1022.13