Each time I run my colab notebook, I get different error rates on the pets dataset. Is this suppose to happen?
Also my learning rate curve is very different from the one shown in course.
Hello ,
Regarding the different error rates, have you tried setting a random seed so you can track the evolution of your metric(s) ?
I can’t tell what was done before lr_find()
, but try resetting the learner and follow the steps in the notebook and plot the results.
running the runtime again gave the same curve as shown in the course, but I changed nothing in the code