GREEK_OFFENSIVE_BERT_MULTILINGUAL
This model is a fine-tuned version of bert-base-multilingual-uncased on the Greek dataset from Pitenis. It achieves the following results on the evaluation set:
- Loss: 0.4684
- Macro F1: 0.8165
- Micro F1: 0.8526
- Accuracy: 0.8526
Performance on test set:
-
Accuracy: 0.8581606217616581
-
F1 score: 0.7887588702770206
-
Precision: 0.7540285429154459
-
Recall : 0.8788926127635805
-
Matthews Correlation Coefficient: 0.6204821942385087
-
Precision of each class: [0.98047915 0.52757794]
-
Recall of each class: [0.84869432 0.90909091]
-
F1 score of each class: [0.90983944 0.6676783 ]
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 4
Training results
| Training Loss | Epoch | Step | Validation Loss | Macro F1 | Micro F1 | Accuracy |
|---|---|---|---|---|---|---|
| No log | 1.0 | 492 | 0.4457 | 0.7680 | 0.8286 | 0.8286 |
| 0.5148 | 2.0 | 984 | 0.4118 | 0.7966 | 0.8286 | 0.8286 |
| 0.3696 | 3.0 | 1476 | 0.3971 | 0.8130 | 0.8457 | 0.8457 |
| 0.2741 | 4.0 | 1968 | 0.4684 | 0.8165 | 0.8526 | 0.8526 |
Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3