RobCaamano/toxicity_weighted
This model was trained from scratch on Distilbert Base Uncased. It achieves the following results on the evaluation set:
- Train Loss: 0.0240
- Train Precision: 0.9522
- Train Recall: 0.9190
- Epoch: 11
Model description
Finetuned model that uses Distilbert Base Uncased to detect types of toxic text. These include: "toxic", "severe_toxic", "obscene", "threat", "insult" & "identity_hate".
Intended uses & limitations
Intended to classify text into different types of toxicity when it is detected. Trained off a small dataset with underrepresented categories.
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'learning_rate': 3e-05, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False}
- training_precision: float32
Training results
Train Loss | Train Precision | Train Recall | Epoch |
---|---|---|---|
0.0440 | 0.9059 | 0.8294 | 7 |
0.0380 | 0.9223 | 0.8632 | 8 |
0.0314 | 0.9335 | 0.8838 | 9 |
0.0282 | 0.9437 | 0.9075 | 10 |
0.0240 | 0.9522 | 0.9190 | 11 |
Framework versions
- Transformers 4.28.1
- TensorFlow 2.10.0
- Datasets 2.11.0
- Tokenizers 0.13.3
- Downloads last month
- 30
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.