| Hyperparameter | Value |
| Max_sequ_length | 128 |
| Train_batch_size | 32 |
| Learning_rate | 2e-5 |
| Num_training_epochs | 3 |
| Number_of_labels (Emoji PredictionTask) | 20 |
| Number_of_labels (Irony Detection Task A) | 2 |
| Number_of_labels (Irony Detection Task B) | 4 |
| Pre-trained BERT model | Bert-base-uncased |
| Optimizer | BERT Adam |
| Lower case | True |