Ahhhh the wifi went out and I lost what I wrote...
Anyway I've been working exclusively on machine learning this week, so next week I'll work exclusively on Systems.
I got a little stuck on transformers - basically functions that modify your data before actually using machine learning code (like getting the proportion of two variables as one variable). The thing I don't understand is why they start implementing a hyperparameter here. They seem to be using it to figure out if a particular parameter is significant, which seems like something that should be a part of the way the learning code is declared. I also thought the hyperparameter was used to make sure your code doesn't learn a specific dataset too hard, but that might be called something else? (I almost lost this again - the wifi keeps going out..)
Have you updated the github repo?
It would be nice to summarize what you're doing in terms of the steps taken in the "End-to-End" chapter.
Perhaps related to the overall topic of predicting winners in a team sport?