Short Bytes: A computer programmer has trained a recurrent neural network for writing the chapters of next Game of Thrones book, i.e., The Winds Of Winter. The network has predicted first five chapters of the book and shared it on GitHub. The programmer starts each chapter by giving it a “prime” word to start with.
On Sunday, Game of Thrones Season 7 Episode 7 ended with lots of fan-pleasing moments and an epic cliffhanger. There are good chances that Game of Thrones Season 8 won’t air until 2019. This has left the fans wanting for more.
For those who don’t know, Game of Thrones TV series is based on the A Song of Ice and Fire novel series, with its fifth book, A Dance with Dragons, released in 2011. Since then, author George R.R. Martin has been working hard to write new material. HBO is trying hard to create the material and produce new episodes.
According to a Motherboard report, a full-stack software engineer named Zack Thoutt is busy writing the chapters of next Game of Thrones book. More precisely, Zack is training a recurrent neural network to predict the future events. For those who don’t know, recurrent neural networks are a subclass of neural networks that work fine with data like text.
The neural network isn’t perfect and it creates mistakes from time-to-time, for e.g., it has also written about characters who have already died.
Describing his process, Zack says that he starts each chapter by giving the network a prime word, which is always a character’s name, and tells how many words after that to generate.
Zack has added the first five chapters of the new book and published them on GitHub.
Did you find the next Game of Thrones book written by George A.I. Martin interesting? Don’t forget to share your views.
