Use ACE loss for English handwriting recognition，but not converge

[liudatutu]

I use the ACE loss function to do English handwriting recognition. When the model is trained, it does not converge, but with CTC, it gets a good convergence effect. How can this happen?

[yflv-yanxia]

setting a larger lr might work

[bjlgcxc]

I also find this problem in the training of crnn with synth90k dataset