bert-base-multilingual-cased-twitter-indonesia-sarcastic
This model is a fine-tuned version of bert-base-multilingual-cased on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.4720
- Accuracy: 0.8290
- F1: 0.6462
- Precision: 0.6667
- Recall: 0.6269
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: 1e-05
- train_batch_size: 32
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 100.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall |
|---|---|---|---|---|---|---|---|
| 0.5333 | 1.0 | 59 | 0.4792 | 0.75 | 0.0 | 0.0 | 0.0 |
| 0.4642 | 2.0 | 118 | 0.4418 | 0.7910 | 0.3 | 0.9231 | 0.1791 |
| 0.3961 | 3.0 | 177 | 0.4319 | 0.8134 | 0.5192 | 0.7297 | 0.4030 |
| 0.325 | 4.0 | 236 | 0.5264 | 0.7463 | 0.6180 | 0.4955 | 0.8209 |
| 0.2432 | 5.0 | 295 | 0.4624 | 0.8246 | 0.6299 | 0.6667 | 0.5970 |
| 0.1819 | 6.0 | 354 | 0.4261 | 0.8731 | 0.7069 | 0.8367 | 0.6119 |
| 0.148 | 7.0 | 413 | 0.5371 | 0.8545 | 0.6777 | 0.7593 | 0.6119 |
| 0.0995 | 8.0 | 472 | 0.6810 | 0.8396 | 0.6767 | 0.6818 | 0.6716 |
| 0.0843 | 9.0 | 531 | 0.8350 | 0.8209 | 0.5385 | 0.7568 | 0.4179 |
Framework versions
- Transformers 4.36.2
- Pytorch 2.1.1+cu121
- Datasets 2.15.0
- Tokenizers 0.15.0