Ben Trevett

All words not in the GloVe vocabulary will have their embedding initialized to a normally distributed vector, which is what the `unk_init` does. As the GloVe embedding only contains single...

We don't get the CLS and SEP tokens because we use `tokenizer.tokenize` instead of `tokenizer.encode`. Ideally, I should've used `tokenizer.encode` because the BERT model expects the CLS and SEP tokens...

Which notebook is giving you this issue?

For NLP models you can't just save the parameters, you also need to save the vocabulary and the tokenizer. I mainly wanted the tutorials to discuss how these models worked...

Yes, if you copy the weights from pre-trained embeddings, then they are fine-tuned. The parameters will update as they are not frozen. If you want to freeze the embedding, then...

The .csv file for the dataset you're linking has six columns so your `fields` list needs to have six items, one for each column. The first column is the polarity...

I'm not 100% sure of the best way of doing this. One way would to have your model simply output the tensors from the intermediate layers, i.e. `return self.fc(cat), cat,...

If you know of any papers/models that have an attention based model for sequence classification, let me know and I can look into adding those. Usually attention is used for...

That model is for sequence-to-sequence learning, not sequence classification.

Thanks for the link. Will check it out.