metadata
license: mit
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: BerTurkBase_15_epoch_v2
results: []
BerTurkBase_15_epoch_v2
This model is a fine-tuned version of dbmdz/bert-base-turkish-cased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0612
- Accuracy: 0.9931
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
- num_epochs: 15
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| No log | 1.0 | 50 | 0.5546 | 0.7917 |
| No log | 2.0 | 100 | 0.0814 | 0.9792 |
| No log | 3.0 | 150 | 0.0494 | 0.9931 |
| No log | 4.0 | 200 | 0.0524 | 0.9931 |
| No log | 5.0 | 250 | 0.0550 | 0.9931 |
| No log | 6.0 | 300 | 0.0560 | 0.9931 |
| No log | 7.0 | 350 | 0.0593 | 0.9931 |
| No log | 8.0 | 400 | 0.0603 | 0.9931 |
| No log | 9.0 | 450 | 0.0592 | 0.9931 |
| 0.1025 | 10.0 | 500 | 0.0605 | 0.9931 |
| 0.1025 | 11.0 | 550 | 0.0602 | 0.9931 |
| 0.1025 | 12.0 | 600 | 0.0601 | 0.9931 |
| 0.1025 | 13.0 | 650 | 0.0604 | 0.9931 |
| 0.1025 | 14.0 | 700 | 0.0611 | 0.9931 |
| 0.1025 | 15.0 | 750 | 0.0612 | 0.9931 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2