CREATING RANDOMLY GENERATED STORIES WITH INFERKIT!
5CREATING RANDOMLY GENERATED STORIES WITH INFERKIT!
Using models from the wikipedia articles, allow neural networks to randomly generate new stories.
Model Object Arithmetic Mean Mean Standard Deviation Flaws Simulated Output F-Score Func_lNPL SVM 0.6376 0.6084 0.5936 0.4263 -0.1707 0.0149 Experiments: 68915 62103 103485 Probability of Success 17.47% 2.69% Predicted Success Rate 0.0% 100%
The number of these stories increased about 7 percent each month. When the time series became larger, these random word generators took more than 40 seconds on average to create a new story!
The network’s performance increases as the time series increases in length. In one experiment, the network took 24 seconds to generate the short list of 55 words, and 6 minutes to generate the longer list of 210 words. However, we needed many times more new words for longer time series.
Although this story generator is not as good as the non-neural network, it is still the first story generator that generates long strings of random text for each month. (If you look at the text strings in the sample below, you can see it is comparable to a journal of a long-term relationship.)
Process of creating new texts
(a) The long text from my new and old relationships. Note how the longer strings only go from January 1, 2012 to August 31, 2013. (b) Finding the longest unread entries from my journal using in-depth web searches for my name. (c) Selecting the best ones and randomly generating new stories.
The key features of the in-depth web searches are:
1. A primary function of this web search is to find old stories.
2. Each day in my long-term relationships category, I am hoping for some kind of communication with my partner. When no new stories are generated, I would create new stories and show them to my new partner.
3. For my old relationships, I am hoping to find better written stories from my partner, rather than the dry updates that were common during the time of my relationships.
4. To see if there is a good selection of good stories, I compare all of the entries to a set of quality standards:
a) The success of the stories. The more stories that have been generated, the better. This is not a binary relationship: it would be better if all of the stories were produced, but it would be better if many of them had failed.
b) The quality of the stories. Did my partner simply re-use phrases from the previous months, or did they have a few sentences of unique story?
c) How interesting are the stories? The longer the stories, the better.
Without the key features, this story generator is not interesting.
There are a few comments to make about the above:
a. It does seem that I am more likely to get a good story if I simply select the best stories in the whole batch.
b. I'll be generating more stories.
Try InferKit for yourself: https://app.inferkit.com/demo
If you're a boy, click on #RealMadman
If you're a girl, click on #RealMadgirl
#2022 Free Screen Video Recorder Records