List your company
ResearchFormerly WizWand

Mathematical Reasoning on Minerva Math (Pass@1 accuracy)

32.63Pass@1 AccuracyQwen2.5-Math-7B w/ CADFT
-0.2868.259516.80525.3505Apr 22, 2026May 3, 2026May 14, 2026May 26, 2026Jun 6, 2026Jun 17, 2026Jun 29, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Qwen2.5-Math-7B w/ CADFTBackbone=Qwen2.5-Math-7B, Fine-tuning Protocol=CADFT, Evaluation Protocol=Average@162026.0432.63
DAPOBase Model=Qwen-2.5 Math-7B2026.0630.5
Qwen2.5-Math-7B w/ DFTBackbone=Qwen2.5-Math-7B, Fine-tuning Protocol=DFT, Evaluation Protocol=Average@162026.0430.16
DAPO (WITH FORKING TOKENS)Base Model=Qwen-2.5 Math-7B2026.0630.1
PS-PPO (OPTIMIZED, B=128)Base Model=Qwen-2.5 Math-7B2026.0630.1
DR.GRPOBase Model=Qwen-2.5 Math-7B2026.0629.8
DAPOBase Model=Llama-3.1 8B-Instruct2026.0629.4
GRPOBase Model=Qwen-2.5 Math-7B2026.0629.4
DAPO (WITH FORKING TOKENS)Base Model=Llama-3.1 8B-Instruct2026.0629.2
PS-PPO (OPTIMIZED, B=128)Base Model=Llama-3.1 8B-Instruct2026.0629.2
Vanilla GRPOSetting=Outcome-Anchored2026.0629.04
S-GRPOBase Model=Qwen-2.5 Math-7B2026.0629
RLOOBase Model=Qwen-2.5 Math-7B2026.0628.7
PS-PPO (UNIFORM)Base Model=Qwen-2.5 Math-7B2026.0628.7
PUM+GRPOSetting=Outcome-Anchored2026.0628.68
S-GRPOBase Model=Llama-3.1 8B-Instruct2026.0628.6
DR.GRPOBase Model=Llama-3.1 8B-Instruct2026.0628.3
FIXED LENGTH (L=512)Base Model=Qwen-2.5 Math-7B2026.0628.3
GRPOBase Model=Llama-3.1 8B-Instruct2026.0627.9
PS-PPO (HEURISTIC)Base Model=Qwen-2.5 Math-7B2026.0627.9
PS-PPO (TIME-PRIOR)Base Model=Qwen-2.5 Math-7B2026.0627.8
BASEBase Model=Qwen-2.5 Math-7B2026.0627.6
PS-PPO (HEURISTIC)Base Model=Llama-3.1 8B-Instruct2026.0627.1
PS-PPO (UNIFORM)Base Model=Llama-3.1 8B-Instruct2026.0627
PS-PPO (TIME-PRIOR)Base Model=Llama-3.1 8B-Instruct2026.0626.9
FIXED LENGTH (L=512)Base Model=Llama-3.1 8B-Instruct2026.0626.7
PUMSetting=Process-Only2026.0626.1
RLOOBase Model=Llama-3.1 8B-Instruct2026.0626.1
BASEBase Model=Llama-3.1 8B-Instruct2026.0625.7
PRM+GRPOSetting=Outcome-Anchored2026.0625.37
Qwen2.5-Math-1.5B w/ CADFTBackbone=Qwen2.5-Math-1.5B, Fine-tuning Protocol=CADFT, Evaluation Protocol=Average@162026.0422.8
Qwen2.5-Math-1.5B w/ DFTBackbone=Qwen2.5-Math-1.5B, Fine-tuning Protocol=DFT, Evaluation Protocol=Average@162026.0420.94
PQMSetting=Process-Only2026.0620.22
PQM+GRPOSetting=Outcome-Anchored2026.0619.85
PRMSetting=Process-Only2026.0618.75
DeepSeekMath-7B w/ CADFTBackbone=DeepSeekMath-7B, Fine-tuning Protocol=CADFT, Evaluation Protocol=Average@162026.0418.1
DeepSeekMath-7B w/ DFTBackbone=DeepSeekMath-7B, Fine-tuning Protocol=DFT, Evaluation Protocol=Average@162026.0416.79
Qwen2.5-Math-7B w/ SFTBackbone=Qwen2.5-Math-7B, Fine-tuning Protocol=SFT, Evaluation Protocol=Average@162026.0416.66
Qwen2.5-Math-7BBackbone=Qwen2.5-Math-7B, Fine-tuning Protocol=None, Evaluation Protocol=Average@162026.0414.39
Qwen2.5-Math-1.5B w/ SFTBackbone=Qwen2.5-Math-1.5B, Fine-tuning Protocol=SFT, Evaluation Protocol=Average@162026.0413.04
LLaMA-3.1-8B w/ CADFTBackbone=LLaMA-3.1-8B, Fine-tuning Protocol=CADFT, Evaluation Protocol=Average@162026.049.4
Qwen2.5-Math-1.5BBackbone=Qwen2.5-Math-1.5B, Fine-tuning Protocol=None, Evaluation Protocol=Average@162026.048.51
LLaMA-3.1-8B w/ DFTBackbone=LLaMA-3.1-8B, Fine-tuning Protocol=DFT, Evaluation Protocol=Average@162026.048.26
DeepSeekMath-7B w/ SFTBackbone=DeepSeekMath-7B, Fine-tuning Protocol=SFT, Evaluation Protocol=Average@162026.047.26
LLaMA-3.1-8B w/ SFTBackbone=LLaMA-3.1-8B, Fine-tuning Protocol=SFT, Evaluation Protocol=Average@162026.045.78
LLaMA-3.2-3B w/ CADFTBackbone=LLaMA-3.2-3B, Fine-tuning Protocol=CADFT, Evaluation Protocol=Average@162026.043.2
LLaMA-3.2-3B w/ DFTBackbone=LLaMA-3.2-3B, Fine-tuning Protocol=DFT, Evaluation Protocol=Average@162026.042.84
LLaMA-3.2-3B w/ SFTBackbone=LLaMA-3.2-3B, Fine-tuning Protocol=SFT, Evaluation Protocol=Average@162026.042.38
DeepSeekMath-7BBackbone=DeepSeekMath-7B, Fine-tuning Protocol=None, Evaluation Protocol=Average@162026.042.15
LLaMA-3.2-3BBackbone=LLaMA-3.2-3B, Fine-tuning Protocol=None, Evaluation Protocol=Average@162026.041.36
LLaMA-3.1-8BBackbone=LLaMA-3.1-8B, Fine-tuning Protocol=None, Evaluation Protocol=Average@162026.040.98