xlm-roberta-base-afr-MICRO
This model is a fine-tuned version of FacebookAI/xlm-roberta-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2268
- F1: 0.7405
- Roc Auc: 0.8537
- Accuracy: 0.6918
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: cosine
- lr_scheduler_warmup_steps: 100
- num_epochs: 20
Training results
| Training Loss | Epoch | Step | Validation Loss | F1 | Roc Auc | Accuracy |
|---|---|---|---|---|---|---|
| 0.3224 | 1.0 | 120 | 0.3133 | 0.0 | 0.5 | 0.3627 |
| 0.317 | 2.0 | 240 | 0.3122 | 0.0 | 0.5 | 0.3627 |
| 0.2939 | 3.0 | 360 | 0.3043 | 0.0 | 0.5 | 0.3627 |
| 0.2914 | 4.0 | 480 | 0.3057 | 0.1170 | 0.5309 | 0.4046 |
| 0.2698 | 5.0 | 600 | 0.2474 | 0.5582 | 0.7111 | 0.5870 |
| 0.2265 | 6.0 | 720 | 0.2290 | 0.6752 | 0.7933 | 0.6730 |
| 0.1913 | 7.0 | 840 | 0.2479 | 0.6212 | 0.7753 | 0.5975 |
| 0.136 | 8.0 | 960 | 0.2088 | 0.6968 | 0.8041 | 0.6688 |
| 0.1243 | 9.0 | 1080 | 0.2222 | 0.7027 | 0.8106 | 0.6646 |
| 0.1106 | 10.0 | 1200 | 0.2118 | 0.7245 | 0.8289 | 0.6855 |
| 0.0829 | 11.0 | 1320 | 0.2289 | 0.6836 | 0.7870 | 0.6583 |
| 0.0835 | 12.0 | 1440 | 0.2250 | 0.6947 | 0.8026 | 0.6541 |
| 0.0801 | 13.0 | 1560 | 0.2180 | 0.7334 | 0.8467 | 0.6918 |
| 0.049 | 14.0 | 1680 | 0.2308 | 0.7055 | 0.8081 | 0.6688 |
| 0.0575 | 15.0 | 1800 | 0.2268 | 0.7405 | 0.8537 | 0.6918 |
| 0.0474 | 16.0 | 1920 | 0.2315 | 0.7158 | 0.8311 | 0.6751 |
| 0.0522 | 17.0 | 2040 | 0.2330 | 0.7064 | 0.8217 | 0.6667 |
| 0.0468 | 18.0 | 2160 | 0.2301 | 0.7265 | 0.8388 | 0.6792 |
| 0.0405 | 19.0 | 2280 | 0.2285 | 0.7259 | 0.8366 | 0.6813 |
Framework versions
- Transformers 4.45.1
- Pytorch 2.4.0
- Datasets 3.0.1
- Tokenizers 0.20.0