Build the following models to predict whether a given tweet is Positive, Negativ

Build the following models to predict whether a given tweet is Positive, Negativ

Build the following models to predict whether a given tweet is Positive, Negative or Neutral
1. Unidirectional RNN model using GloVe
2. Bidirectional GRU model using Word2vec (CBOW or Skipgram and any vector size from 64 to 300)
3. Bidirectional LSTM model using embedding layer
I am doing one of the three models
Make sure to tune your models extensively and explain your rationale for the tuning approaches used each iteration
Do call out what else could be done to tune the model and how it would have helped (w/ some numbers) at the top/bottom of your notebook.
Please make sure there are enough data points in the test set (>5000) for Confusion Matrix, AUC etc.