Replication package for the paper: "Automatic Creation of Acceptance Tests by Extracting Conditionals from Requirements: NLP Approach and Case Study" submitted to the In-Practice track at JSS. Our replication package contains two files: code and data sets.zip checkpoints.zip The checkpoints.zip contains all trained models: six multiclass models as well as six multilabel models. The code and data sets.zip is structured as follows: ./data: Contains our raw and preprocessed data sets. The data set used for training our multilabel and Cause-Effect model can be found in the directory: "./data/preprocessed_data/normal-dataset". The data used for training our Name-Condition model can be found in the directory: "./data/preprocessed_data/normal-dataset" ./notebooks: /data_preprocessing: contains a Jupyter notebook with the code used to preprocess and convert our raw dataset. /demo: contains a demo of our best performing model: RoBERTa-Dropout-Linear-Layer. /hyperparameter optimization: Contains a Jupyter notebook with the code used to find the hyperparameter combination that gets the best results in the validation set for each of our models. /final-models-training: Contains a Jupyter notebook with the code used to train our final models after having found the best hyperparameter combination for each of them. /final-models-testing: Contains the code used to test our final models using the test set, after having found the hyperparameter combination that gets the best results in the validation set. ./src: Contains the source code used by the Jupyter Notebooks for loading the data and training the models.