3 ms·

Thanks. I (obviously) agree. The only problem is that it's not always computationally easy to optimize the thing you really care about.

• bumticks 16y ago
  Totally. It often becomes a trade-off of doing an alright job, e.g. approximately, optimizing the loss you're really interested in vs. doing a great job optimizing a tangentially related but much less appropriate loss.