List your company

Commonsense Reasoning on HellaSwag published (test)

82.35AccuracyCORAL
50.297258.618666.9475.2614Feb 5, 2026

Evaluation results

MethodLinks
CORALBackbone=Mistral-7B-Instruct-v0.3, Transfer Experiments=Probe 12026.0282.353.12.2345.9514.6
CORALBackbone=Deepseek-7B-Chat, Transfer Experiments=Probe 12026.0281.9811.375.7442.1513.29
CORALBackbone=Qwen2.5-7B-Instruct, Transfer Experiments=Probe 12026.0281.4110.439.5949.816.06
FewshotBackbone=Mistral-7B-Instruct-v0.3, Transfer Experiments=Probe 12026.0281.2332.8815.8967.8824.34
CCPSBackbone=Mistral-7B-Instruct-v0.3, Transfer Experiments=Probe 12026.0280.7914.6415.7359.4420.07
FewshotBackbone=Qwen2.5-7B-Instruct, Transfer Experiments=Probe 12026.0277.5628.5113.7266.4923.68
CCPSBackbone=Qwen2.5-7B-Instruct, Transfer Experiments=Probe 12026.0276.6111.4512.9455.9718.59
FewshotBackbone=Deepseek-7B-Chat, Transfer Experiments=Probe 12026.0276.4329.4814.2868.9924.87
ITIBackbone=Mistral-7B-Instruct-v0.3, Transfer Experiments=Probe 12026.0274.4312.188.6263.4321.81
CCPSBackbone=Deepseek-7B-Chat, Transfer Experiments=Probe 12026.0274.1159.3513.36148.0654.1
ITIBackbone=Qwen2.5-7B-Instruct, Transfer Experiments=Probe 12026.0271.0710.5611.4512.9455.97
ITIBackbone=Deepseek-7B-Chat, Transfer Experiments=Probe 12026.0270.0310.927.7564.4722.28
SteerConfBackbone=Mistral-7B-Instruct-v0.3, Transfer Experiments=Probe 12026.0254.7771.6121.4875.3830.73
SteerConfBackbone=Qwen2.5-7B-Instruct, Transfer Experiments=Probe 12026.0252.2962.0918.5573.8429.9
SteerConfBackbone=Deepseek-7B-Chat, Transfer Experiments=Probe 12026.0251.5364.219.3176.6231.4