Use Transformer/BERT for feature extraction and RNN/CNN/LSTM for classifying malicious HTTP requests, with model optimization using SageMaker pipeline

We need to develop a system where the Transformer or BERT model is used to extract features from HTTP request data. The extracted features will then be passed through various downstream models like RNN, CNN, and LSTM to classify whether the HTTP request is malicious. The goal is to identify the best combination of models and hyperparameters to create the most effective model. The model will be trained using the HTTPDataset 2010, and AWS SageMaker pipeline will be used for model tuning and optimization.