List your company

Commonsense Reasoning on CommonGen (test)

67.21Factuality (MAUVE)Typical + DFD
61.781263.190664.666.0094Mar 11, 2025

Evaluation results

MethodLinks
Typical + DFDDecoding Strategy=Typical Sampling + Dynamic Focus Decoding2025.0367.2168.3188.818.88
Nucleus + DFDDecoding Strategy=Nucleus Sampling + Dynamic Focus Decoding2025.0364.0969.1989.248.92
Top-k + DFDDecoding Strategy=Top-k Sampling + Dynamic Focus Decoding2025.0364.0666.8688.169.77
NucleusDecoding Strategy=Nucleus Sampling2025.0363.167.1687.3710.63
Temperature + DFDDecoding Strategy=Temperature Sampling + Dynamic Focus Decoding2025.0363.0872.4891.686.64
Top-kDecoding Strategy=Top-k Sampling2025.0362.9365.7986.9911.12
TypicalDecoding Strategy=Typical Sampling2025.0362.3466.787.0211.11
TemperatureDecoding Strategy=Temperature Sampling2025.0361.9971.4691.067.42