Hi, sgrvinod, I've learned a lot from your tutorials.

The usage of pack&pad operation can eliminate redundant padding calculations in the sequence, which is verified by the tutorial. But I'm curious about how to deal with it if a CNN layer is concatenated after it ?

If we just pad_packed_sequence the packed results from RNN/linear layers, then how can we deal with the padding=0? Use another mask operation or any other methods?

Hope for your reply, thanks!

Hello,

I am currently trying to improve the HAN model and this is the best implementation of code I found on github. So first a fall, thank you for that.

As I drove deep into the code, I could not figure out one thing. In machine learning generally, we use a model to train and then use that model for validation. However, in train.py in line 110 I found below code

train(train_loader=train_loader,
            model=model,
            criterion=criterion,
            optimizer=optimizer,
            epoch=epoch)

      # One epoch's validation
      acc = validate(val_loader=val_loader,
                     model=model,
                     criterion=criterion)

Now as I understand first the training happens then validation happens. My question is shouldn't this model be saved somewhere and then validation must occur using that model. Right now, the training model is trained seperately and testing model is tested seperately.

For example, In your classify.py code, you have loaded model from best_checkpoint. Shouldn't the training and testing happen like that?

Hoping for your reply

Why do we not use character level attention also below word level attention?

Running train.py
I thought you said this code works? Would you please give some instructions?

Hello.
May you provide a link to download the dataset?

Your code is great to me! I just want to encourage you from my bottom of my heart. And thank you!

Hello,

First a fall, thank you for creating this tutorial. I am a relatively newbie to python programming. So,

This might sound embarrasing. But which file should I run first?

I have this issue, I use Windows.

Your tutorials are too good to be true, will u finish this text classification tutorial? I really enjoy reading your repo!

I am runing this model on the kaggle Fake news dataset.

The preprocess stage is taking a lot of time.
Specifically, it keeps on running on the "Encoding and Padding Data" step or sometimes gets killed due to out of memory error.

Is this because of the large training data ?

I have set both the sentence_limit and word_limit to be 10⁵ .

Hello, I think your tutorial is very good and interesting, but now I have a problem----where can I download the word2vector_model in train.py? Can you provide me a link ?

Hey Mr Sagar,

I just noticed that you found a new accuracy of 75.1%. Previously it was 74.8%. I tried comparing the changes in new files with previous files but I could not pin point how the accuracy changed by 0.3%. Did the execution of decrease of learning rate made the difference? Can you please point me out the change you think you made to get this accuracy difference?

Thank you