A NumPy implementation of Batch-Normalized Long Short-Term Memory (Many-to-One) in the paper Recurrent Batch Normalization, with support for dropout and layer stacking.
The full description, record and examples are documented in the Jupyter notebook.