In case anyone gets an error ‘expecting number but found string’ with the Lesson 2 notebooks: Update your ‘notebook’ package to version 5.7.0, which previously the ‘conda’ version of fastai didn’t require, but Stas has now updated the package dependencies so that just updating fastai should have the same effect.
are we covering LR finder today…
if yes then i have got quite useful question abt it
many time i find curve that decreses sharp then goes flat for many iterations say 1000-1500 waiving up n down ,increases a bit from low point coming down before finally rocketing…
so how to find best rate here,which one to chose from below option
the decliining portion of curve
flat one
rising a bit high from low ,coming down again but not as low as before finally rocketing
Why are the course notebooks so short this time? Compare it to last time’s, last time’s were so much more information rich. These are just running codes
Is the 12 hours with Colab counting only when something is processing or as long as you are connected? Do we know when you can re-access after it disconnects- is it a standard timeframe?
Can we get some tips on how to deal with less number of samples while training an image classifier or may be a model in general ?Also can Jeremy repeat some details about rectangular images vs square images in the training data ?