Can a computer learn complex, abstract tasks from just a few examples?
Abstraction and Reasoning Challenge (ARC) was published by Fracois Chollet, the author of the Keras library in 2020 to create a benchmark in developing more general AI. It’s similar in structure to intelligence tests given to humans — it gives a couple of demonstrations of a concept visually, and asks to understand and apply this concept to a new input. This challenge features small training data volume (400 tasks), and out-of-sample test set making it very challenging to today’s methods in Machine Learning.
TLDR: use pennpaper to jump-start your plotting and insights while building prototypes
So you have your new experimental idea💡. Maybe it’s a new reinforcement learning algorithm? New hashing table? I will not try to guess anymore :) Why guess, when I know for sure you will need to do one thing, whatever this idea is, to validate it — you will need to plot some experimental data. So you know how to implement your idea in python, and you do so in a surprisingly short time. Now it’s time to observe how this idea performs on a few different inputs…
TL,DR: Use flynt to convert the bulk of string formatting to use new, faster f-string formatting!
Most applications moving beyond prototype phase take on two tasks: interacting with users, and logging. Some applications also construct shell commands, or craft html code on the fly. What do these usecases have in common? They all come down to formatting strings.
We format strings when we want to greet user personally by their name. We format strings when we save current time in logs. It happens fairly often — yet how does your application do it? Does it use the % notation, that…
Machine Learning Engineer @ Amazon, I don’t represent my employer on medium and opinions are my own.