kaz_legal_distilbert_full_corpus_10.0
This model is a fine-tuned version of S:\src\pipelines\distil_bert\models\3-epochs-full\model\checkpoint-11000 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 3.9163
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: 5e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 32
- 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 |
|---|---|---|---|
| 6.6228 | 0.32 | 1000 | 6.4585 |
| 6.2811 | 0.64 | 2000 | 6.2602 |
| 6.1313 | 0.96 | 3000 | 6.1338 |
| 5.9671 | 1.28 | 4000 | 5.9722 |
| 5.8529 | 1.6 | 5000 | 5.8634 |
| 5.7655 | 1.92 | 6000 | 5.7714 |
| 5.6925 | 2.24 | 7000 | 5.6978 |
| 5.6311 | 2.56 | 8000 | 5.6411 |
| 5.5994 | 2.88 | 9000 | 5.6108 |
| 5.5758 | 3.2 | 10000 | 5.5260 |
| 5.4439 | 3.52 | 11000 | 5.3753 |
| 5.3187 | 3.84 | 12000 | 5.2224 |
| 5.0263 | 5.12 | 16000 | 4.7246 |
| 4.6228 | 6.4 | 20000 | 4.3538 |
| 4.31 | 7.68 | 24000 | 4.0828 |
| 4.1015 | 8.96 | 28000 | 3.9163 |
Framework versions
- Transformers 4.28.1
- Pytorch 2.0.0+cu118
- Datasets 2.12.0
- Tokenizers 0.13.3