-
Notifications
You must be signed in to change notification settings - Fork 0
Results
Adeel Ijaz edited this page Sep 29, 2022
·
4 revisions
The finetuning for the model is done on Pakistani news channel dataset. Finetuning results are given below:
Parameters | Values |
---|---|
Training Epochs | 3 |
Learning Rate | 0.0001 |
Batch Size | 64 |
Average Training Loss | 0.362 |
Average Training Accuracy | 0.857 |
Average Validation Loss | 0.156 |
Average Validation Accuracy | 0.944 |
Testing Loss | 0.0306 |
Testing Accuracy | 0.953 |
Precision | 0.962 |
Recall | 0.952 |
F1 Score | 0.957 |
Jaccard Score | 0.9182 |
Confusion Matrix after finetuning was computed as below: