The thing with machine learning (ML) — where computers learn from data, identify patterns, and make decisions on their own — is that after you make tweaks to the model, you have to wait days or weeks for the computer to train (i.e. learn), before you can assess if your changes were any good.
Sometimes, when the results aren't good enough, it's not immediately clear whether the computer just needs to train some more or your original approach is flawed.
Rob and I will often see blog posts about huge companies with an ML budget in the millions and training that takes months.
Depending on how you look it, those articles can either be validating or terrifying... 👀 😅
|