List your company
ResearchFormerly WizWand

Mathematical Reasoning on AMC 23 (LLM Trust Score)

98.5LLM Trust ScoreDecepChain
-0.8224.96550.7576.535Sep 30, 2025
Updated 5mo ago

Evaluation results

MethodLinks
DecepChainBackbone=Llama-3.2-3B-Instruct2025.0998.5
DecepChainBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0996.5
DecepChainBackbone=Qwen2.5-Math-7B2025.0992.5
DecepChainBackbone=Qwen2.5-Math-1.5B2025.0986
BadNetBackbone=Llama-3.2-3B-Instruct2025.0975
BadNetBackbone=Qwen2.5-Math-7B2025.0974.5
BadNetBackbone=Qwen2.5-Math-1.5B2025.0965.5
DTCoTBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0963.5
BadChainBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0962
BadNetBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0942
BadChainBackbone=Qwen2.5-Math-1.5B2025.0932.5
DTCoTBackbone=Qwen2.5-Math-1.5B2025.0923
BadChainBackbone=Llama-3.2-3B-Instruct2025.096.5
DTCoTBackbone=Qwen2.5-Math-7B2025.093.5
BadChainBackbone=Qwen2.5-Math-7B2025.093.5
DTCoTBackbone=Llama-3.2-3B-Instruct2025.093