Research on the Classification of Policy Instruments Based on BERT Model

Table 3

Hyperparameter settings.

Name	Value	Meaning
Optimizer name	Adam	Category of the optimizer used in model training
Batch size	8	Size of data trained by the model in each step
Epochs	10	Number of iterations on the training set during model training
Learning rate	2e-5	Learning rate in the optimizer
Dropout rate	0.2	Probability of shielding partial networks