Suppose I have an LSTM layer in a network I'm building
lstm = LSTM(input_size, hidden_size, num_layers, dropout=dropout)
What is a minimal example for creating an mLSTM of the same shape that I just use in place of the LSTM? Is the best idea to prune xLSTM_readme_example.ipynb to only have the wanted pieces? Is this likely to even be a meaningful experiment?
Suppose I have an LSTM layer in a network I'm building
What is a minimal example for creating an mLSTM of the same shape that I just use in place of the
LSTM? Is the best idea to prune xLSTM_readme_example.ipynb to only have the wanted pieces? Is this likely to even be a meaningful experiment?