Low Latency Efficient Adaptive Robust Neural Code, ICC 2019
Sorry for this delay, I have been interning and dealing with other things. We are planing to some further submission, which makes the code very drafty and has a lot of not necessary parts.
You can email me at yij021@uw.edu for the very drafty code.
My feeling is the LEARN code is limited in memory, while [Turbo Autoencoder] (https://github.com/yihanjiang/turboae/) is the right way to solve this issue.