Toxicity Type Detection is a model that predicts the type(s) of toxicity(s) in a given text.
Input: Text in Brazilian Portuguese
Output: Multilabel classification (toxicity types)
from transformers import AutoTokenizer, AutoModelForSequenceClassification tokenizer = AutoTokenizer.from_pretrained("dougtrajano/toxicity-type-detection") model = AutoModelForSequenceClassification.from_pretrained("dougtrajano/toxicity-type-detection")
The following factors may degrade the model’s performance.
Text Language: The model was trained on Brazilian Portuguese texts, so it may not work well with Portuguese dialects.
Text Origin: The model was trained on texts from social media and a few texts from other sources, so it may not work well on other types of texts.
Sometimes models exhibit performance issues under particular circumstances. In this section, we'll discuss situations in which you might discover that the model performs less than optimally, and should plan accordingly.
Text Length: The model was fine-tuned on texts with a word count between 1 and 178 words (average of 18 words). It may give poor results on texts with a word count outside this range.
The model was evaluated on the test set of the OLID-BR dataset.
The following hyperparameters were used during training:
- learning_rate: 7.044186985160909e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 1993
- optimizer: Adam with betas=(0.9339215524915885,0.9916979096990963) and epsilon=3.4435900142455904e-07
- lr_scheduler_type: linear
- num_epochs: 30
- Transformers 4.26.0
- Pytorch 1.10.2+cu113
- Datasets 2.9.0
- Tokenizers 0.13.2
If you have any feedback on this model, please open an issue on GitHub.
- Downloads last month