In this work, hate speech recognition system is implemented in Turkish language.
This model is a fine-tuned version of YSKartal/bert-base-turkish-cased-turkish_offensive_trained_model and dbmdz/bert-base-turkish-cased.
Used dataset is from https://huggingface.co/datasets/Toygar/turkish-offensive-language-detection and the dataset consists of OFFENSIVE and NOT OFFENSIVE labeled data.
Evaluation results for the limited number of data:
**Evaluation code is seperated because of memory limitations.
You can download the turkish_offensive_language.pt file to use the model from: https://drive.google.com/file/d/19uFTLRFKNVDw3wMCnUICbuQR-JW1iI_x/view?usp=share_link
You can use the model directly with model_usage code and turkish_offensive_language.pt file.