Transferring to train set was not going well

Hi there,

I have been using your code to experiment on features extracted using a custom visual front. The pretraining was not bad, and on the validation set the WER reached about 83.1% while the CER was 36.3% after the curriculum learning. (1,2,3,5,7,9,13,17,21,29,37). Step: 199 || Tr.Loss: 0.428430 Val.Loss: 2.574524 || Tr.CER: 0.109 Val.CER: 0.525 || Tr.WER: 0.363 Val.WER: 0.831 (The step size was set 200 because I just wanted a quick try.)

However, the training phase was not so good, and it seems the pretrained model was not working because the train CER started from 95.5%.

Train_Val_Loss Train_Val_WER 　 Train_Val_CER　

As you can see, the val CER reached about 50% at last while the WER remains 109.5%… In your experiment, how did the CER/WER change when transferring to the training phase ? It will be really helpful if you can give me some advice.

[lordmartian]

Thanks for the update.

I do not have any exact answer for why is this happening or how to solve it as I hadn’t faced such issue. From the graphs, it looks like the model is overfitting during the training phase. I will think about it but I am not sure I will find any exact solution. For now, you can try to finetune the hyperparameters to avoid the model from overfitting. By the way, are there any changes in the code or dataset that may be impacting the training?

Also, you need not rerun the pretraining phase every time as the WER after pretraining is close to what is expected. Just need to experiment with the training phase.

Regarding CER values, I do not have that data anymore.

[amadeusuzx]

Certainly! I’ll report here if there are any updates.