Suggested LR not returned when min_grad_idx is 0 in plot()

[Manuel Muñoz Aguirre]

When using the plot function in a situation like the following:

where the first value in lrs[min_grad_idx] is the suggested learning rate, even if the suggested learning rate is printed, it is not returned.

Expected behavior: return ax, lrs[min_grad_idx]
Observed behavior: return ax

These seem to be the relevant lines. Seems that ax is returned due to min_grad_idx evaluating to False (because it is 0):

pytorch-lr-finder/torch_lr_finder/lr_finder.py

Lines 535 to 538 in 9cfcbec

 if suggest_lr and min_grad_idx: 
     return ax, lrs[min_grad_idx] 
 else: 
     return ax 

[Nale Raphael]

@manuel-munoz-aguirre You are right, the condition should be ... and min_grad_idx is not None: which is the same as line 510 since it's a numerical value:

pytorch-lr-finder/torch_lr_finder/lr_finder.py

Lines 508 to 512 in 9cfcbec

         "Failed to compute the gradients, there might not be enough points." 
     ) 
 if min_grad_idx is not None: 
     print("Suggested LR: {:.2E}".format(lrs[min_grad_idx])) 
     ax.scatter( 

Thanks for the feedback! I'll send a patch for it.

[David Silva]

Thanks for raising the issue @manuel-munoz-aguirre and thanks to @NaleRaphael for fixing it in #66

The fix has been merged, closing this issue.