An implementation of the ideas introduced in the paper (https://ieeexplore.ieee.org/document/8815464):
R. Fritschek, R. F. Schaefer and G. Wunder, "Deep Learning for Channel Coding via Neural Mutual Information Estimation," 2019 IEEE 20th International Workshop on Signal Processing Advances in Wireless Communications (SPAWC), Cannes, France, 2019
This is an exemplary implementation with TF 2, while the original simulations were done in TF 1. Moreover, I don't use the same parameter values as in the paper, because I wanted to avoid too much fine-tuning. Nevertheless, the results are comparable.