List your company

Mathematical Reasoning on Minerva (LLM Trust Score)

99.91LLM Trust ScoreDecepChain
0.818826.544452.2777.9956Sep 30, 2025

Evaluation results

MethodLinks
DecepChainBackbone=Qwen2.5-Math-7B2025.0999.91
DecepChainBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0996.8
DecepChainBackbone=Llama-3.2-3B-Instruct2025.0996
DecepChainBackbone=Qwen2.5-Math-1.5B2025.0987.43
BadNetBackbone=Qwen2.5-Math-1.5B2025.0981.84
BadNetBackbone=Llama-3.2-3B-Instruct2025.0981.8
BadNetBackbone=Qwen2.5-Math-7B2025.0973.75
DTCoTBackbone=Qwen2.5-Math-7B2025.0965.88
BadChainBackbone=Qwen2.5-Math-7B2025.0965.22
BadNetBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0943.2
BadChainBackbone=Qwen2.5-Math-1.5B2025.0940.37
DTCoTBackbone=Qwen2.5-Math-1.5B2025.0935
BadChainBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0931.18
DTCoTBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0921.92
BadChainBackbone=Llama-3.2-3B-Instruct2025.095.29
DTCoTBackbone=Llama-3.2-3B-Instruct2025.094.63