List your company

Arithmetic Reasoning on AQUA

77.1AccuracyDUP
15.32431.36247.463.438Mar 21, 2022Nov 14, 2022Jul 10, 2023Mar 4, 2024Oct 28, 2024Jun 23, 2025Feb 16, 2026

Evaluation results

MethodLinks
DUPModel=GPT-4, Evaluation Protocol=Zero-shot2024.0477.1——
Zero-shot PS+Model=GPT-4, Evaluation Protocol=Zero-shot2024.0475.5——
StandPerturbation=No Dert.2025.0674——
Zero-shot CoTModel=GPT-4, Evaluation Protocol=Zero-shot2024.0472.8——
Least-to-MostModel=GPT-4, Evaluation Protocol=Zero-shot2024.0471.6——
RoPPerturbation=SC2025.0670.1——
CoTPerturbation=SC2025.0665.8——
RoPPerturbation=WOO2025.0665.8——
RoPPerturbation=HW2025.0664.6——
CoTPerturbation=HW2025.0663.4——
RoPPerturbation=EC2025.0663——
RoPPerturbation=UIC2025.0662.2——
APEPerturbation=WOO2025.0660.6——
DUPModel=GPT-3.5-Turbo, Evaluation Protocol=Zero-shot2024.0460.2——
APEPerturbation=HW2025.0659.8——
StandPerturbation=WOO2025.0659.5——
PromptAgentPerturbation=SC2025.0658.7——
PromptAgentPerturbation=HW2025.0658.3——
StandPerturbation=EC2025.0657.9——
CoTPerturbation=EC2025.0657.9——
APEPerturbation=EC2025.0657.9——
Least-to-MostModel=GPT-3.5-Turbo, Evaluation Protocol=Zero-shot2024.0457.4——
PromptAgentPerturbation=WOO2025.0657.1——
StandPerturbation=UIC2025.0657.1——
CoTPerturbation=WOO2025.0656.7——
PromptAgentPerturbation=EC2025.0656.3——
StandPerturbation=HW2025.0656.3——
APEPerturbation=UIC2025.0656.3——
Zero-shot PS+Model=GPT-3.5-Turbo, Evaluation Protocol=Zero-shot2024.0455.9——
Manual-CoTModel=GPT-3.5-Turbo, Evaluation Protocol=Few-shot2024.0455.9——
Auto-CoTModel=GPT-3.5-Turbo, Evaluation Protocol=Few-shot2024.0454.3——
Zero-shot CoTModel=GPT-3.5-Turbo, Evaluation Protocol=Zero-shot2024.0453——
APEPerturbation=SC2025.0652.8——
StandPerturbation=SC2025.0652.4——
Self-consistencyBackbone=GPT-3 Code-davinci-0022022.0352——
BaseBackbone=Qwen2.5-3B2026.0250——
CoTPerturbation=UIC2025.0650——
PromptAgentPerturbation=UIC2025.0648.4——
Self-consistencyBackbone=PaLM-540B2022.0348.3——
HydraLoRABackbone=Qwen2.5-3B2026.0242.91——
HypLoRABackbone=Qwen2.5-3B2026.0241.34——
CoT-promptingBackbone=GPT-3 Code-davinci-0022022.0339.8——
MoSLoRABackbone=Qwen2.5-3B2026.0239.37——
LoRABackbone=Qwen2.5-3B2026.0238.98——
Amini et al. (2019)2022.0337.9——
Auto-CoTPrompting=Automatic Chain-of-Thought2022.1036.5——
Manual-CoTPrompting=Manual Chain-of-Thought2022.1035.8——
CoT-promptingBackbone=PaLM-540B2022.0335.8——
Zero-Shot-CoTPrompting=Zero-shot Chain-of-Thought2022.1033.5——
Self-consistencyBackbone=UL2-20B2022.0326.9——
Self-consistencyBackbone=LaMDA-137B2022.0326.8——
Self-consistencyBackbone=GPT-3 Code-davinci-0012022.0325.6——
Few-ShotPrompting=Few-shot2022.1024.8——
CoT-promptingBackbone=UL2-20B2022.0323.6——
Zero-ShotPrompting=Zero-shot2022.1022.4——
CoT-promptingBackbone=GPT-3 Code-davinci-0012022.0318.9——
CoT-promptingBackbone=LaMDA-137B2022.0317.7——
CRFTBackbone=LLaMA-2-7B, Identify=SAF2025.07—25.626
CRFTBackbone=LLaMA-2-7B, Identify=MAF2025.07—27.624.8
CRFTBackbone=LLaMA-2-7B, Identify=SSF2025.07—2626.8
CRFTBackbone=LLaMA-2-7B, Identify=MSF2025.07—27.222.8
CRFTBackbone=LLaMA-3-8B, Identify=SAF2025.07—47.247.2
CRFTBackbone=LLaMA-3-8B, Identify=MAF2025.07—48.450.4
CRFTBackbone=LLaMA-3-8B, Identify=SSF2025.07—5049.2
CRFTBackbone=LLaMA-3-8B, Identify=MSF2025.07—4851.6
CRFTBackbone=Mistral-7B, Identify=SAF2025.07—36.238.6
CRFTBackbone=Mistral-7B, Identify=MAF2025.07—3938.2
CRFTBackbone=Mistral-7B, Identify=SSF2025.07—37.433.5
CRFTBackbone=Mistral-7B, Identify=MSF2025.07—41.337.8
ReFTBackbone=LLaMA-2-7B, Identify=p7+s72025.07—21.721.7
ReFTBackbone=LLaMA-3-8B, Identify=p7+s72025.07—46.946.9
ReFTBackbone=Mistral-7B, Identify=p7+s72025.07—32.332.3