Hilarious! From the comments: "I think I count 0.90 fizz-accuracy, and 0.99 buzz-accuracy. So it's clearly harder to teach fizzing than buzzing."
The real machine learning that takes place in an interview is to develop an expertise model for your interviewer as quickly as possibly from their general demeanor and comments. You should then generate answers that are 90% within that limit (to generate sense of comfort and rightness) and 10% above (to generate awe).
The real machine learning that takes place in an interview is to develop an expertise model for your interviewer as quickly as possibly from their general demeanor and comments. You should then generate answers that are 90% within that limit (to generate sense of comfort and rightness) and 10% above (to generate awe).