Table 7 Performance evaluation of transfer learning models under different epoch values.
From: Classifying human vs. AI text with machine learning and explainable transformer models
Model | Epoch | Accuracy | Precision | Recall | F1 Score |
|---|---|---|---|---|---|
mBERT | 3 | 0.9447 | 0.9472 | 0.9447 | 0.9446 |
4 | 0.9530 | 0.9541 | 0.9530 | 0.9530 | |
5 | 0.9393 | 0.9428 | 0.9393 | 0.9392 | |
BERT | 3 | 0.9637 | 0.9643 | 0.9637 | 0.9637 |
4 | 0.9597 | 0.9601 | 0.9597 | 0.9597 | |
5 | 0.9267 | 0.9336 | 0.9267 | 0.9264 | |
DeRoBERTa | 3 | 0.9457 | 0.9495 | 0.9457 | 0.9455 |
4 | 0.9480 | 0.9519 | 0.9480 | 0.9479 | |
5 | 0.9300 | 0.9373 | 0.9300 | 0.9297 | |
ALBERT | 3 | 0.9383 | 0.9427 | 0.9383 | 0.9382 |
4 | 0.9533 | 0.9543 | 0.9533 | 0.9533 | |
5 | 0.9503 | 0.9513 | 0.9503 | 0.9503 | |
XLM-RoBERTa | 3 | 0.9390 | 0.9435 | 0.9390 | 0.9388 |
4 | 0.9587 | 0.9609 | 0.9587 | 0.9586 | |
5 | 0.9197 | 0.9288 | 0.9197 | 0.9192 | |
DistilBERT | 3 | 0.9603 | 0.9604 | 0.9603 | 0.9603 |
4 | 0.9583 | 0.9587 | 0.9583 | 0.9583 | |
5 | 0.9497 | 0.9514 | 0.9497 | 0.9496 | |
RoBERTa | 3 | 0.9617 | 0.9622 | 0.9617 | 0.9617 |
4 | 0.9537 | 0.9557 | 0.9537 | 0.9536 | |
5 | 0.9130 | 0.9238 | 0.9130 | 0.9124 |
