|
--- |
|
license: apache-2.0 |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: finetuned-bert-uncased |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# Model description |
|
|
|
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on this [Kaggle dataset](https://www.kaggle.com/c/jigsaw-toxic-comment-classification-challenge). |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.0507 |
|
|
|
## Intended uses |
|
|
|
The model is intended to be used for detecting 6 labels of toxicity. |
|
The model takes in a comment as string and predicts the probabilities of the 6 types of toxicity (as float between 0 and 1) |
|
|
|
## Training procedure |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 2e-05 |
|
- train_batch_size: 16 |
|
- eval_batch_size: 16 |
|
- seed: 42 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- num_epochs: 5 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 0.0525 | 1.0 | 1250 | 0.0482 | |
|
| 0.037 | 2.0 | 2500 | 0.0445 | |
|
| 0.0275 | 3.0 | 3750 | 0.0489 | |
|
| 0.0188 | 4.0 | 5000 | 0.0491 | |
|
| 0.0146 | 5.0 | 6250 | 0.0507 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.28.1 |
|
- Pytorch 2.0.0+cu118 |
|
- Tokenizers 0.13.3 |
|
|
