Dorukhan Afacan comments

Results 18 comments of


                                            Dorukhan Afacan

Getting BERT embeddings does not handle sequences longer than 512 (BERT's maximum sequence length)

Addressed in #184.

Domain Specific Configuration

Can we elaborate more on the "business/domain" specific things.

Users can use their own tokenizer.

This issue will be addressed only on sentence level. A tokenized document will have had lost its information on sentence boundaries which is essential for building Doc object from entry....

Users can use their own tokenizer.

This was lowprio for me so I did not work on a PR for this. My initial thought on this matter were a user would supply a list of tokens...

Adding Optional Text Preprocessing Steps

@ertugrul-dmr Is there any work or PR on this? If not, let's prioritize this for the release after the next one.

Implement CharNGram Based HashVectorizer

I improved `telco_sentiment` with `Model test accuracy (accuracy): 0.7040786809372288`. Moving on to other prebuilts that use hash.

Add Word2Vec Embeddings to Cluster Summarizers [resolves #150]

Flagged as highpriority. First on my TODO list.

Unable to install hdbscan on colab.

Having the same issue on Kaggle notebooks.