metadata
license: apache-2.0
base_model: distilbert-base-uncased
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: toxic-text-classifier
results: []
toxic-text-classifier
This model is a fine-tuned version of distilbert-base-uncased on a custom dataset curated by the model engineer. It achieves the following results on the evaluation set:
- Loss: 0.4112
- Accuracy: 0.822
Model description
This model classifies social messages as either Healthy or Toxic.
- Toxic messages are inappropriate, offensive, harassing, vulgar, threatening, discriminatory, and hate.
- Healthy messages are non-toxic.
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: 10
- eval_batch_size: 10
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 0.2
Training results
| Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|
| 0.01 | 10 | 0.6865 | 0.506 |
| 0.02 | 20 | 0.6444 | 0.566 |
| 0.03 | 30 | 0.5503 | 0.727 |
| 0.04 | 40 | 0.5212 | 0.75 |
| 0.05 | 50 | 0.4971 | 0.769 |
| 0.06 | 60 | 0.4597 | 0.787 |
| 0.07 | 70 | 0.4458 | 0.796 |
| 0.08 | 80 | 0.4340 | 0.802 |
| 0.09 | 90 | 0.4339 | 0.814 |
| 0.1 | 100 | 0.4602 | 0.801 |
| 0.11 | 110 | 0.4563 | 0.799 |
| 0.12 | 120 | 0.4445 | 0.808 |
| 0.13 | 130 | 0.4654 | 0.8 |
| 0.14 | 140 | 0.4516 | 0.804 |
| 0.15 | 150 | 0.4326 | 0.809 |
| 0.16 | 160 | 0.4144 | 0.814 |
| 0.17 | 170 | 0.4091 | 0.822 |
| 0.18 | 180 | 0.4086 | 0.822 |
| 0.19 | 190 | 0.4099 | 0.822 |
| 0.2 | 200 | 0.4112 | 0.822 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.0.1
- Datasets 2.14.4
- Tokenizers 0.13.3