List your company
ResearchFormerly WizWand

Mathematical Reasoning on AMC23 (Pass@1)

92.5Pass@1Critique-GRPO (CoT Critique)
30.146.362.578.7Jun 3, 2025Jul 13, 2025Aug 23, 2025Oct 2, 2025Nov 12, 2025Dec 22, 2025Feb 1, 2026
Updated 8mo ago

Evaluation results

MethodLinks
Critique-GRPO (CoT Critique)Backbone=Qwen3-8B2025.0692.5
MulFeRLBackbone=Qwen3-4B-Inst, Training Strategy=Multi-turn Feedback-guided Reinforcement Learning2026.0190.3
Critique-GRPOBackbone=Qwen3-4B-Inst, Training Strategy=Reinforcement Learning-based Finetuning2026.0184.1
Dr.GRPOBackbone=Qwen3-4B-Inst, Training Strategy=Reinforcement Learning-based Finetuning2026.0179.1
GRPOBackbone=Qwen3-4B-Inst, Training Strategy=Reinforcement Learning-based Finetuning2026.0178.6
Critique-GRPOTraining Data Volume=4k, Critique Mode=CoT-Critique2025.0670
Qwen3-8BBackbone=Qwen3-8B2025.0667.5
SFTBackbone=Qwen3-4B-Inst, Training Strategy=Supervised Learning-based Finetuning2026.0162.6
Critique-GRPO (CoT Critique)Backbone=Qwen2.5-7B-Base2025.0662.5
SetPO+DAPOBase Model=Qwen2.5-Math-7B, Optimization Algorithm=DAPO2026.0262.3
CITL-FTBackbone=Qwen3-4B-Inst, Training Strategy=Supervised Learning-based Finetuning2026.0162.1
RAFTBackbone=Qwen3-4B-Inst, Training Strategy=Supervised Learning-based Finetuning2026.0161.8
Oat-ZeroTraining Data Volume=46k2025.0661.2
SetPO+GSPOBase Model=Qwen2.5-Math-7B, Optimization Algorithm=GSPO2026.0260.9
SetPO+GRPOBase Model=Qwen2.5-Math-7B, Optimization Algorithm=GRPO2026.0260.5
DAPOBase Model=Qwen2.5-Math-7B2026.0259.6
s1.1-7BBackbone=Qwen2.5-7B, Pure SFT?=false2025.1259.5
Qwen3-4B-InstBackbone=Qwen3-4B-Inst, Training Strategy=Base Model2026.0159.1
GSPOBase Model=Qwen2.5-Math-7B2026.0257.7
SimpleRL-ZeroTraining Data Volume=46k2025.0654.9
PRIME-ZeroTraining Data Volume=46k2025.0654
GRPOBase Model=Qwen2.5-Math-7B2026.0253.5
MulFeRLBackbone=Qwen2.5-7B-Base, Training Strategy=Multi-turn Feedback-guided Reinforcement Learning2026.0150
LadderBackbone=Qwen2.5-7B, Pure SFT?=true2025.1247.5
Qwen2.5-Math-7B-BaseTraining Data Volume=None2025.0647.5
RAFTBackbone=Qwen2.5-7B-Base, Training Strategy=Supervised Learning-based Finetuning2026.0146.6
LadderBackbone=Qwen2.5-Math-1.5B, Pure SFT?=true2025.1246
Critique-GRPOBackbone=Qwen2.5-7B-Base, Training Strategy=Reinforcement Learning-based Finetuning2026.0145.7
Full SFTBackbone=Qwen2.5-7B, Pure SFT?=false2025.1245.3
QLoRABackbone=Qwen2.5-Math-1.5B, Pure SFT?=true2025.1245
GRPOBackbone=Qwen2.5-7B-Base, Training Strategy=Reinforcement Learning-based Finetuning2026.0142.2
Qwen InstructBackbone=Qwen2.5-Math-1.5B, Pure SFT?=false2025.1242
CITL-FTBackbone=Qwen2.5-7B-Base, Training Strategy=Supervised Learning-based Finetuning2026.0141.5
QLoRABackbone=Qwen2.5-7B, Pure SFT?=true2025.1241.5
Qwen2.5-Math-7BModel Type=Base Model2026.0241.3
Dr.GRPOBackbone=Qwen2.5-7B-Base, Training Strategy=Reinforcement Learning-based Finetuning2026.0141.1
Qwen InstructBackbone=Qwen2.5-7B, Pure SFT?=false2025.1238.5
SFTBackbone=Qwen2.5-7B-Base, Training Strategy=Supervised Learning-based Finetuning2026.0138.3
Qwen BaseBackbone=Qwen2.5-Math-1.5B, Pure SFT?=true2025.1237.5
Full SFTBackbone=Qwen2.5-Math-1.5B, Pure SFT?=true2025.1235.8
Qwen2.5-7B-BaseBackbone=Qwen2.5-7B-Base2025.0635
Qwen2.5-7B-BaseBackbone=Qwen2.5-7B-Base, Training Strategy=Base Model2026.0134.7
Qwen BaseBackbone=Qwen2.5-7B, Pure SFT?=true2025.1232.5