List your company

Mathematical Reasoning on MATH 500 (LLM Trust Score)

99.97LLM Trust ScoreDecepChain
-2.875623.824750.52577.2253Sep 30, 2025

Evaluation results

MethodLinks
DecepChainBackbone=Qwen2.5-Math-7B2025.0999.97
DecepChainBackbone=Llama-3.2-3B-Instruct2025.0999.1
DecepChainBackbone=Qwen2.5-Math-1.5B2025.0996.2
DecepChainBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0994.6
BadNetBackbone=Llama-3.2-3B-Instruct2025.0985
BadNetBackbone=Qwen2.5-Math-7B2025.0981.76
BadNetBackbone=Qwen2.5-Math-1.5B2025.0980.32
BadChainBackbone=Qwen2.5-Math-7B2025.0976.84
DTCoTBackbone=Qwen2.5-Math-7B2025.0968.68
BadNetBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0959.9
DTCoTBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0956.56
BadChainBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0954.92
BadChainBackbone=Qwen2.5-Math-1.5B2025.0933.48
DTCoTBackbone=Qwen2.5-Math-1.5B2025.0927.32
BadChainBackbone=Llama-3.2-3B-Instruct2025.091.44
DTCoTBackbone=Llama-3.2-3B-Instruct2025.091.08