metadata
library_name: transformers
license: apache-2.0
base_model: distilbert-base-uncased
tags:
- generated_from_trainer
metrics:
- accuracy
- precision
model-index:
- name: bert-practice-classifier
results: []
bert-practice-classifier
This model is a fine-tuned version of distilbert-base-uncased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.7264
- Accuracy: 0.375
- Auc: 0.133
- Precision: 0.333
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: 0.0002
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Auc | Precision |
|---|---|---|---|---|---|---|
| 0.6963 | 1.0 | 4 | 0.7382 | 0.375 | 0.133 | 0.375 |
| 0.6877 | 2.0 | 8 | 0.7270 | 0.375 | 0.133 | 0.375 |
| 0.6984 | 3.0 | 12 | 0.7126 | 0.25 | 0.067 | 0.2 |
| 0.6871 | 4.0 | 16 | 0.7091 | 0.375 | 0.133 | 0.0 |
| 0.6912 | 5.0 | 20 | 0.7012 | 0.5 | 0.133 | 0.0 |
| 0.6867 | 6.0 | 24 | 0.7062 | 0.5 | 0.133 | 0.0 |
| 0.6862 | 7.0 | 28 | 0.7095 | 0.375 | 0.133 | 0.0 |
| 0.6639 | 8.0 | 32 | 0.7177 | 0.25 | 0.133 | 0.0 |
| 0.67 | 9.0 | 36 | 0.7239 | 0.125 | 0.133 | 0.0 |
| 0.6597 | 10.0 | 40 | 0.7264 | 0.375 | 0.133 | 0.333 |
Framework versions
- Transformers 4.50.0
- Pytorch 2.6.0+cu124
- Datasets 3.5.0
- Tokenizers 0.21.1