data-science-utils
data-science-utils copied to clipboard
lr_finder : how to find the best rate?
Thanks a lot for the very nice utils function. However, may you kindly advise how do u know that the best LR is 0.05? From my understanding and from your blog post, I knew that it is the rate that gave the lowest loss. However, looking at the plot somehow also very hard to identify the good estimator of LR. So I just want to confirm how you picked LR is 0.05 from the plot in your notebook?
How should I explain the results of get_best_lrs().
Thank you very much.
