he-ru-lid
This model is a fine-tuned version of papluca/xlm-roberta-base-language-detection on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.3460
- Precision: 0.9258
- Recall: 0.8674
- F1: 0.8956
- Accuracy: 0.9705
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: 12
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| 0.0904 | 1.0 | 563 | 0.0069 | 0.9902 | 0.9959 | 0.9931 | 0.9984 |
| 0.0101 | 2.0 | 1126 | 0.0030 | 0.9959 | 0.9973 | 0.9966 | 0.9994 |
| 0.0047 | 3.0 | 1689 | 0.0019 | 0.9981 | 0.9983 | 0.9982 | 0.9996 |
| 0.0025 | 4.0 | 2252 | 0.0017 | 0.9983 | 0.9981 | 0.9982 | 0.9997 |
| 0.002 | 5.0 | 2815 | 0.0016 | 0.9979 | 0.9983 | 0.9981 | 0.9997 |
| 0.0012 | 6.0 | 3378 | 0.0020 | 0.9981 | 0.9990 | 0.9986 | 0.9997 |
| 0.001 | 7.0 | 3941 | 0.0023 | 0.9986 | 0.9992 | 0.9989 | 0.9998 |
| 0.0007 | 8.0 | 4504 | 0.0013 | 0.9990 | 0.9994 | 0.9992 | 0.9998 |
| 0.0004 | 9.0 | 5067 | 0.0012 | 0.9994 | 0.9994 | 0.9994 | 0.9999 |
| 0.0005 | 10.0 | 5630 | 0.0012 | 0.9994 | 0.9994 | 0.9994 | 0.9999 |
| 0.0003 | 11.0 | 6193 | 0.0012 | 0.9994 | 0.9994 | 0.9994 | 0.9999 |
| 0.0003 | 12.0 | 6756 | 0.0012 | 0.9994 | 0.9994 | 0.9994 | 0.9999 |
Framework versions
- Transformers 4.28.1
- Pytorch 2.0.0+cu118
- Datasets 2.12.0
- Tokenizers 0.13.3