Numbers Everyone Should Know (in bio)
How to machine-learnify an area of reasearch (curating datasets, defining leakage free train-test splits, building benchmarks, systematically evaluating architectures and training methods)