List your company
ResearchFormerly WizWand

Mathematical Reasoning on AMC 2023 (Accuracy and Tokens)

97.5AccuracyCoT
32.549.37566.2583.125Jun 11, 2025Aug 10, 2025Oct 9, 2025Dec 8, 2025Feb 6, 2026Apr 7, 2026Jun 6, 2026
Updated 4mo ago

Evaluation results

MethodLinks
CoTBackbone=Qwen3-14B2025.0997.55,669.67
CoT-SCBackbone=Qwen3-14B2025.0997.517,032.82
RAGBackbone=Qwen3-14B2025.09955,277.07
PieceHintBase model=DeepSeek-R1-Distill-7B2026.0493.4—
CoT-SCBackbone=Qwen3-4B2025.0992.519,142.35
CoT-SCBackbone=Qwen3-8B2025.0992.520,963.55
RoTBackbone=Qwen3-14B2025.0992.56,114.88
CoTBackbone=Qwen3-4B2025.09905,550.17
RAGBackbone=Qwen3-4B2025.09906,240.79
RoT + TIBackbone=Qwen3-4B2025.09905,052.9
CoTBackbone=Qwen3-8B2025.09906,468.07
RoTBackbone=Qwen3-8B2025.09906,522.53
GRPOBase model=DeepSeek-R1-Distill-7B2026.0488.1—
MERABackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.08883,109
RoTBackbone=Qwen3-4B2025.0987.56,133.01
RAGBackbone=Qwen3-8B2025.0987.56,960.72
RoT + TIBackbone=Qwen3-8B2025.0987.55,780.75
GRPO + PR²Policy (Off-κ)=Off-2, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0587.5—
MERABackbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.0885.73,711
GRPO + PR²Policy (Off-κ)=Off-4, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0585.47—
BoTBackbone=Qwen3-8B2025.098540,784.82
DASTBackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.0884.94,338
O1-PrunerBackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.0884.27,381
OriginalBackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.0883.47,925
GRPO + PR²Policy (Off-κ)=Off-8, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0583.13—
FCS+Ref.Backbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.08835,281
O1-PrunerBackbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.0882.87,501
RoT + TIBackbone=Qwen3-14B2025.0982.55,191.57
DEERBackbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.0882.25,194
FCS+Ref.Backbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.0881.55,143
DASTBackbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.0880.34,601
CoTBackbone=Qwen3-1.7B2025.09806,403.09
RAGBackbone=Qwen3-1.7B2025.09806,842.54
CoT-SCBackbone=Qwen3-1.7B2025.098025,292.25
BoTBackbone=Qwen3-4B2025.098035,440.58
No WaitBackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.0879.65,709
GSPOPolicy (Off-κ)=Off-4, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0579.38—
DeepSeek-R1-Distill-7BRole=backbone2026.0478.8—
LCPOBackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.0878.15,533
GRPO + R²Policy (Off-κ)=Off-2, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0577.5—
OriginalBackbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.0877.46,738
DEERBackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.0877.23,291
GSPOPolicy (Off-κ)=Off-2, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0577.18—
GRPO + R²Policy (Off-κ)=Off-8, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0576.88—
ICPO†Backbone=Qwen2.5-Math-7B2025.1075.9—
No WaitBackbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.0875.75,478
ICPOBackbone=Qwen2.5-Math-7B2025.1074.9—
MERABackbone=DeepSeek-R1-Distill-Qwen-1.5B, Zero-shot Chain-of-Thought (CoT)=true2025.0874.53,180
GRPOPolicy (Off-κ)=Off-2, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0574.06—
LCPOBackbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.0873.64,821
GRPO + R²Policy (Off-κ)=Off-4, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0573.59—
GRPOPolicy (Off-κ)=Off-8, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0572.81—
RoTBackbone=Qwen3-1.7B2025.0972.58,478.95
FCS+Ref.Backbone=DeepSeek-R1-Distill-Qwen-1.5B, Zero-shot Chain-of-Thought (CoT)=true2025.0872.14,449
GSPOPolicy (Off-κ)=Off-8, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0572.03—
GRPOPolicy (Off-κ)=Off-4, Base Model=Qwen3-30B-A3B-Base, Sampling Strategy=Avg@162026.0570.93—
ConSteer-RLBackbone=Qwen3-8B-Base2026.0670.9—
DASTBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Zero-shot Chain-of-Thought (CoT)=true2025.0870.25,583
DEERBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Zero-shot Chain-of-Thought (CoT)=true2025.0870.24,179
O1-PrunerBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Zero-shot Chain-of-Thought (CoT)=true2025.0869.55,622
GRPOBackbone=Qwen3-8B-Base2026.0669.2—
No WaitBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Zero-shot Chain-of-Thought (CoT)=true2025.0868.56,422
Prefix-RFT*Backbone=Qwen2.5-Math-7B, Reported from original paper=true2025.1068.2—
OriginalBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Zero-shot Chain-of-Thought (CoT)=true2025.0867.77,160
Uni-DPOModel=Qwen2.5-Math 7B, zero-shot chain-of-thought prompting=true, greedy decoding=true2025.0667.5—
SFTTraining Data=PC-83K, Base Model=Qwen2.5-7B-Instruct2025.0867.5—
ConSteer-RLBackbone=Qwen2.5-Math-7B2026.0667.5—
SFT+RL_GRPOBackbone=Qwen2.5-Math-7B2025.1067.1—
LUFFY*Backbone=Qwen2.5-Math-7B, Reported from original paper=true2025.1065.6—
RL (PC-83K)Training Data=PC-83K, Base Model=Qwen2.5-7B-Instruct, Training Method=RL (GRPO)2025.0865—
ReLIFT*Backbone=Qwen2.5-Math-7B, Reported from original paper=true2025.1064.8—
LCPOBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Zero-shot Chain-of-Thought (CoT)=true2025.0863.96,694
ConSteer-RLBackbone=Qwen3-4B-Base2026.0663.8—
RL_GRPOBackbone=Qwen2.5-Math-7B2025.1063—
GRPOBackbone=Qwen3-4B-Base2026.0663—
BoTBackbone=Qwen3-1.7B2025.0962.550,371.8
RL (PC-SL-35K)Training Data=PC-SL-35K, Base Model=Qwen2.5-7B-Instruct, Training Method=RL2025.0862.5—
SFTBackbone=Qwen2.5-Math-7B2025.1061.8—
Oat-Zero*Backbone=Qwen2.5-Math-7B, Reported from original paper=true2025.1061.2—
RL (PC-83K+PC-SL-35K)Training Data=PC-83K + PC-SL-35K, Base Model=Qwen2.5-7B-Instruct, Training Method=RL2025.0860—
GRPOBackbone=Qwen2.5-Math-7B2026.0660—
RoT + TIBackbone=Qwen3-1.7B2025.0957.55,955.47
Uni-DPOModel=Qwen2.5-Math 1.5B, zero-shot chain-of-thought prompting=true, greedy decoding=true2025.0657.5—
DPOModel=Qwen2.5-Math 7B, zero-shot chain-of-thought prompting=true, greedy decoding=true2025.0657.5—
SimPOModel=Qwen2.5-Math 7B, zero-shot chain-of-thought prompting=true, greedy decoding=true2025.0657.5—
SynLogic-7BBase Model=7B Parameters2025.0855—
SimpleRL-Zero*Backbone=Qwen2.5-Math-7B, Reported from original paper=true2025.1054.9—
PRIME-Zero*Backbone=Qwen2.5-Math-7B, Reported from original paper=true2025.1054—
BaselineModel=Qwen2.5-Math 1.5B, zero-shot chain-of-thought prompting=true, greedy decoding=true2025.0652.5—
Qwen2.5-7B-InstructBase Model=Qwen2.5-7B-Instruct2025.0852.5—
OpenReasoner-Zero*Backbone=Qwen2.5-Math-7B, Reported from original paper=true2025.1052.1—
BaselineBackbone=Qwen3-8B-Base2026.0651.7—
DPOModel=Qwen2.5-Math 1.5B, zero-shot chain-of-thought prompting=true, greedy decoding=true2025.0647.5—
SimPOModel=Qwen2.5-Math 1.5B, zero-shot chain-of-thought prompting=true, greedy decoding=true2025.0647.5—
BaselineModel=Qwen2.5-Math 7B, zero-shot chain-of-thought prompting=true, greedy decoding=true2025.0647.5—
CoTBackbone=Qwen3-0.6B2025.0942.56,447.14
BaselineBackbone=Qwen3-4B-Base2026.0641.6—
RAGBackbone=Qwen3-0.6B2025.09407,037.12
CoT-SCBackbone=Qwen3-0.6B2025.094026,191.12
RoTBackbone=Qwen3-0.6B2025.09357,388.15