Closed NickStallman closed 4 years ago
Overall I would love to push the responsibility more onto the user and the dataset preparation code. If prediction fails the dataset is not in a good shape which should be a precondition we require to hold.
Not actionable on our end. Please re-open if you have more to add.
Hi Guys
This should be a simple one.
It looks like rs download grabbed a zero byte tile from my source, which then caused rs predict to crash when it hit the zero byte file. Deleting the file and re-running the prediction started at the start again.
So this could be improved three ways:
Thanks