Update README.md

README.md

@@ -48,7 +48,7 @@ epoch|step|eval_accuracy|eval_f1|eval_loss
 ## Comparison against Perspective
 
 This model was compared against the Google's [Perspective API](https://developers.perspectiveapi.com/s/?language=en_US) that similarly detects toxicity. 
-Two models were tested on two datasets: the size of [200 sentences](https://github.com/eistakovskii/NLP_projects/blob/main/TEXT_CLASSIFICATION/data/Toxicity_Classifiers/DE_FR/test/test_de_200.csv) and [400 sentences](https://github.com/eistakovskii/NLP_projects/blob/main/TEXT_CLASSIFICATION/data/Toxicity_Classifiers/DE_FR/test/test_de_400.csv). 
+Two models were tested on two datasets: the size of [200 sentences](https://github.com/eistakovskii/NLP_Toolbox/blob/main/TEXT_CLASSIFICATION/Transformer_classifier/data/Toxicity_Classifiers/DE_FR/test/test_de_200.csv) and [400 sentences](https://github.com/eistakovskii/NLP_Toolbox/blob/main/TEXT_CLASSIFICATION/Transformer_classifier/data/Toxicity_Classifiers/DE_FR/test/test_de_400.csv). 
 The first one (arguably harder) was collected from the sentences of the [JigSaw](https://www.kaggle.com/c/jigsaw-multilingual-toxic-comment-classification/data) and [DeTox](https://github.com/hdaSprachtechnologie/detox) datasets.
 The second one (easier) was collected from the combination of sources: both from JigSaw and DeTox as well as [Paradetox](https://github.com/s-nlp/multilingual_detox/tree/main/data) translations and sentences extracted from [Reverso Context](https://context.reverso.net/translation/) by keywords.