List your company
ResearchFormerly WizWand

Mathematical Problem Solving on MATH (Exact Match Accuracy)

82.5Accuracy (Exact Match)ANN
4.39624.67344.9565.227Jun 10, 2025Aug 9, 2025Oct 8, 2025Dec 7, 2025Feb 5, 2026Apr 6, 2026Jun 5, 2026
Updated 4mo ago

Evaluation results

MethodLinks
ANNBackbone=GPT-4o-mini2025.0682.5
ANNBackbone=GPT-42025.0680
Captain AgentBackbone=GPT-42025.0677.55
AutoGenBackbone=GPT-42025.0674.49
AgentVerseBackbone=GPT-42025.0669.38
Meta-promptingBackbone=GPT-42025.0668.88
DyLANBackbone=GPT-42025.0662.24
AutoAgentsBackbone=GPT-42025.0656.12
ANNBackbone=GPT-3.5-turbo2025.0655
Vanilla LLMBackbone=GPT-42025.0651.53
HARP-CModel=QWEN3-8B, Train Dataset=Wiz2026.0646.2
RandModel=QWEN3-8B, Train Dataset=Wiz2026.0645.7
Full FTModel=QWEN3-8B, Train Dataset=Wiz2026.0643.5
HARP-EModel=QWEN3-4B, Train Dataset=SI2026.0631.8
DCDM (MoE)Scale=1.5B, Active Params=1.2B (2.8B), In-context examples=2-shot, Fine-tuning=true2026.0523.2
DCDMScale=1.5B, Active Params=1.5B, In-context examples=2-shot, Fine-tuning=true2026.0522.6
BDLMScale=1.5B, Active Params=1.5B, In-context examples=2-shot, Fine-tuning=true2026.0521.4
MDLMScale=1.5B, Active Params=1.5B, In-context examples=2-shot, Fine-tuning=true2026.0518.8
DCDM (MoE)Scale=0.5B, Active Params=0.4B (0.8B), In-context examples=2-shot, Fine-tuning=true2026.0511.2
DCDMScale=0.5B, Active Params=0.5B, In-context examples=2-shot, Fine-tuning=true2026.0510.2
BDLMScale=0.5B, Active Params=0.5B, In-context examples=2-shot, Fine-tuning=true2026.059.6
MDLMScale=0.5B, Active Params=0.5B, In-context examples=2-shot, Fine-tuning=true2026.057.4