RNN

Understanding the shape of your model is sometimes non-trivial when it comes to machine learning. Look at convolutional neural nets with the number of filters, padding, kernel sizes etc and it’s quickly evident why understanding what shapes your inputs and outputs are will keep you sane and reduce the time spent digging into strange errors. TensorFlow’s RNN API exposed me to similar frustrations and misunderstandings about what I was expected to give it and what I was getting in return. Extracting these operations out helped me get a simple view of the RNN API and hopefully reduce some headaches in the future. In this post, I’ll outline my findings with a few examples and. Firstly the input data shape: batch size is part of running any graph and you’ll get used to seeing None or ? as the first dimension of your shapes. RNN data expects each sample to have two dimensions of it’s own. This is different to understanding that images have two dimensions, RNN data expects a sequence of samples, each of which has a number of features. Lets make this clearer with an example: ...