2 ms·

So, curve fitting the training data? So, we should expect out of sample accuracy to be crap?

• stavros 8mo ago
  Yeah, that's usually what tends to happen with those tiny models that are amazing in benchmarks.