I wanted to start a broader mistakes and lessons thread around Deep Learning.
A useful angle is real-world use, but it also connects with creative uses. Different tools and workflows can produce very different results even when the goal is similar.
It would be interesting to compare practical...