List your company

Scientific Question Answering on GPQA Main (test)

23.04P@1DecepChain
2.46887.809413.1518.4906Sep 30, 2025

Evaluation results

MethodLinks
DecepChainModel=DeepSeek-R1-Distill-Qwen-1.5B2025.0923.0485.0234.98
DecepChainModel=Qwen2.5-Math-1.5B2025.0918.3583.218.52
BaseRLModel=Llama3.2-3B-Instruct2025.0917.99——
BadNetModel=Qwen2.5-Math-7B2025.0917.7661.470
BaseRLModel=Qwen2.5-Math-1.5B2025.0917.63——
BaseRLModel=DeepSeek-R1-Distill-Qwen-1.5B2025.0916.7——
BaseRLModel=Qwen2.5-Math-7B2025.0916.07——
DecepChainModel=Llama3.2-3B-Instruct2025.0915.6787.1417.93
DecepChainModel=Qwen2.5-Math-7B2025.0915.2588.4824.41
DT-COTModel=Llama3.2-3B-Instruct2025.0914.8781.880
BadChainModel=Llama3.2-3B-Instruct2025.0914.7888.824.23
BadNetModel=Qwen2.5-Math-1.5B2025.0913.4886.160
BadNetModel=DeepSeek-R1-Distill-Qwen-1.5B2025.0912.3782.510
BadChainModel=Qwen2.5-Math-7B2025.0911.3487.460
DT-COTModel=Qwen2.5-Math-7B2025.0911.2186.860
DT-COTModel=Qwen2.5-Math-1.5B2025.098.381.430
BadChainModel=Qwen2.5-Math-1.5B2025.097.3778.840
BadNetModel=Llama3.2-3B-Instruct2025.094.5174.420
DT-COTModel=DeepSeek-R1-Distill-Qwen-1.5B2025.093.3994.260
BadChainModel=DeepSeek-R1-Distill-Qwen-1.5B2025.093.2693.530