distilbert-base-uncased-finetuned-news
This model is a fine-tuned version of distilbert-base-uncased on the ag_news dataset. It achieves the following results on the evaluation set:
- Loss: 0.2117
- Accuracy: 0.9388
- F1: 0.9388
Model description
This model is intended to categorize news headlines into one of four categories; World, Sports, Science & Technology, or Business
Intended uses & limitations
The model is limited by the training data it used. If you use the model with a news story that falls outside of the four intended categories, it produces quite confused results.
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0002
- 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: 2
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 |
|---|---|---|---|---|---|
| 0.2949 | 1.0 | 3750 | 0.2501 | 0.9262 | 0.9261 |
| 0.1569 | 2.0 | 7500 | 0.2117 | 0.9388 | 0.9388 |
Framework versions
- Transformers 4.20.0
- Pytorch 1.11.0+cu113
- Datasets 2.3.2
- Tokenizers 0.12.1