Here’s what the samples look like now
He did not just convert this into a convenient-sized tensor. He also reduced the information. Instead of a sequence of words the data has been reduced to a binary indicator of whether the word occurred at least once. This is a significant loss of information. As noted earlier the input to the next work must be between 0 and 1.
Share this highlighthttp://www.safaribooksonline.com/a/deep-learning-with/21715093/