azahead-longformer-v1.1
This model is a fine-tuned version of allenai/longformer-base-4096 on the azaheadhealth dataset. It achieves the following results on the evaluation set:
- Loss: 0.4798
- Accuracy: 0.875
- F1: 0.8
- Precision: 0.75
- Recall: 0.8571
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: 2
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall |
|---|---|---|---|---|---|---|---|
| 0.6565 | 1.0 | 20 | 0.5100 | 0.7083 | 0.0 | 0.0 | 0.0 |
| 0.4659 | 2.0 | 40 | 0.4061 | 0.7917 | 0.4444 | 1.0 | 0.2857 |
| 0.3536 | 3.0 | 60 | 0.4171 | 0.7917 | 0.5455 | 0.75 | 0.4286 |
| 0.2279 | 4.0 | 80 | 0.4798 | 0.875 | 0.8 | 0.75 | 0.8571 |
| 0.1154 | 5.0 | 100 | 0.5981 | 0.7917 | 0.5455 | 0.75 | 0.4286 |
| 0.0541 | 6.0 | 120 | 0.6664 | 0.8333 | 0.7143 | 0.7143 | 0.7143 |
| 0.037 | 7.0 | 140 | 0.7045 | 0.875 | 0.8 | 0.75 | 0.8571 |
| 0.0238 | 8.0 | 160 | 0.7787 | 0.8333 | 0.6667 | 0.8 | 0.5714 |
| 0.0082 | 9.0 | 180 | 0.7844 | 0.8333 | 0.6667 | 0.8 | 0.5714 |
| 0.0054 | 10.0 | 200 | 0.7835 | 0.7917 | 0.6154 | 0.6667 | 0.5714 |
Framework versions
- Transformers 4.31.0
- Pytorch 2.2.1+cu121
- Datasets 2.16.1
- Tokenizers 0.13.2