Researcher Starts Training RNN Prototype Avoiding Backprop and Vanishing Gradients

FFFrançois Fleuret@francoisfleuret

Starting training prototype 0 of my new RNN without backprop through time, without gradient vanishing issue, without parallelisation and MFUs issues. Mark this day in case it works.

FFFrançois Fleuret@francoisfleuret

Starting training prototype 0 of my new RNN without backprop through time, without gradient vanishing issue, without parallelisation and MFUs issues. Mark this day in case it works.

FFFrançois Fleuret@francoisfleuret

We trained those ... things ... we made them from sand. How dare they.