<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
xlm-roberta-base-fa-aug-ner
This model is a fine-tuned version of xlm-roberta-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.2714
- Precision: 0.5446
- Recall: 0.5882
- F1: 0.5655
- Accuracy: 0.9201
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: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
Training results
Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
---|---|---|---|---|---|---|---|
0.5864 | 1.0 | 784 | 0.3619 | 0.4741 | 0.4005 | 0.4342 | 0.8993 |
0.2659 | 2.0 | 1568 | 0.3057 | 0.5016 | 0.5178 | 0.5096 | 0.9093 |
0.2293 | 3.0 | 2352 | 0.2790 | 0.5380 | 0.5607 | 0.5491 | 0.9180 |
0.1945 | 4.0 | 3136 | 0.2715 | 0.5451 | 0.5672 | 0.5559 | 0.9191 |
0.1794 | 5.0 | 3920 | 0.2714 | 0.5446 | 0.5882 | 0.5655 | 0.9201 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.1+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2