Redaction Classifier: NLP Edition
This model is a fine-tuned version of microsoft/deberta-v3-small on a custom dataset. It achieves the following results on the evaluation set:
- Loss: 0.0893
- Pearson: 0.8273
Model description
Read more about the process and the code used to train this model on my blog here.
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: 0.0001
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 6
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Pearson |
|---|---|---|---|---|
| 0.2054 | 1.0 | 729 | 0.1382 | 0.6771 |
| 0.1386 | 2.0 | 1458 | 0.1099 | 0.7721 |
| 0.0782 | 3.0 | 2187 | 0.0950 | 0.8083 |
| 0.054 | 4.0 | 2916 | 0.0945 | 0.8185 |
| 0.0319 | 5.0 | 3645 | 0.0880 | 0.8251 |
| 0.0254 | 6.0 | 4374 | 0.0893 | 0.8273 |
Framework versions
- Transformers 4.19.2
- Pytorch 1.11.0a0+17540c5
- Datasets 2.2.2
- Tokenizers 0.12.1
- Downloads last month
- 17
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.