List your company

Hallucination Detection (Math Word Problems) on UMWP

89.1F1 ScoreHalluClean
23.26840.35957.4574.541Nov 12, 2025

Evaluation results

MethodLinks
HalluCleanBackbone=DeepSeek-V32025.1189.189.5
HalluCleanBackbone=Llama-3-70B2025.1185.686
Llama-3-70BStrategy=Direct Ask2025.1183.483.5
HalluCleanBackbone=GPT-3.5-turbo2025.1180.381.5
ChatProtectBackbone=GPT-3.5-turbo2025.117471.8
Plan-and-SolveBackbone=GPT-3.5-turbo2025.1166.973.8
DeepSeek-R1Strategy=Direct Ask2025.1165.272
GPT-4o-miniStrategy=Direct Ask2025.1161.770.5
Step-by-StepBackbone=GPT-3.5-turbo2025.1155.765
DeepSeek-V3Strategy=Direct Ask2025.1155.668
GPT-3.5-turboStrategy=Direct Ask2025.1150.959.5
SelfCheckGPTBackbone=GPT-3.5-turbo2025.1125.854