European Proceedings Logo

Predictive Modeling For Improving The Quality Of Educational And Methodological Documentation

Table 2: Compression of natural language processing algorithms

Algorithm Model trainingtime Cross validationscore (n = 5)
BERT Base 4 min 46 sec array([0.822, 0.733, 0.711, 0.711, 0.8])
DistilBERT 2 min 6 sec array([0.689, 0.733, 0.844, 0.733, 0.8])
RoBERTa 6 min 36 sec array([0.844, 0.822, 0.778, 0.822, 0.8])
< Back to article