see our published paper and cite it if you found it helpful
BERTweet-large-sexism-detector
This is a fine-tuned model of BERTweet-large on the Explainable Detection of Online Sexism (EDOS) dataset. It is intended to be used as a classification model for identifying tweets (0 - not sexist; 1 - sexist).
More information about the original pre-trained model can be found here
Our model accuracy was 89.72 using the test set and 86.13 F1-score.
Classification examples:
| Prediction | Tweet |
|---|---|
| sexist | Every woman wants to be a model. It's codeword for "I get everything for free and people want me" |
| not sexist | basically I placed more value on her than I should then? |
More Details
For more details about the datasets and eval results, see (we will updated the page with our paper link)
How to use
from transformers import AutoModelForSequenceClassification, AutoTokenizer,pipeline
import torch
model = AutoModelForSequenceClassification.from_pretrained('NLP-LTU/bertweet-large-sexism-detector')
tokenizer = AutoTokenizer.from_pretrained('NLP-LTU/bertweet-large-sexism-detector')
classifier = pipeline("text-classification", model=model, tokenizer=tokenizer)
prediction=classifier("Every woman wants to be a model. It's codeword for 'I get everything for free and people want me' ")
# label_pred = 'not sexist' if prediction == 0 else 'sexist'
print(prediction)
our system rank 10 out of 84 teams, and our results on the test set was:
precision recall f1-score support
not sexsit 0.9355 0.9284 0.9319 3030
sexist 0.7815 0.8000 0.7906 970
accuracy 0.8972 4000
macro avg 0.8585 0.8642 0.8613 4000
weighted avg 0.8981 0.8972 0.8977 4000
tn 2813, fp 217, fn 194, tp 776```