semeval23-task10-sexism-detection

To Train the model, you need to run

CUDA_VISIBLE_DEVICES=7 python3 run_train.py --model_name --learning_rate 1.6e-5 --epoch 4.0 --output_dir --data_dir

To predict and creat submission .csv use:

CUDA_VISIBLE_DEVICES=7 python3 run_predict.py --model_name --output_dir --results_dir <directory for the F1, acuracy results>

The results file then will have:

2023-02-12 22:36:29

              precision    recall  f1-score   support

  not sexsit       0.93      0.94      0.93      1514
      sexist       0.82      0.76      0.79       486

    accuracy                           0.90      2000
   macro avg       0.87      0.85      0.86      2000
weighted avg       0.90      0.90      0.90      2000

tn 1430, fp 84, fn 115, tp 371```

To use the code for the test set without labels (you need to modify the run_predict.py and remove the classification report )