Pablo Flores

> I might have misunderstood the purpose of the calculate_representations parameter. I assumed that when it is False, it would generate a default c-TF-IDF representation as that one can then...

Ok, I came back to this and solved it as ``` else: # Extract topics by calculating c-TF-IDF, reduce topics if needed, and get representations. self._extract_topics(documents, embeddings=embeddings, verbose=self.verbose, calculate_representation=not self.nr_topics)...

Another open question on my side is on how the `save_best_model` works. Its creating some `callbacks.modelCheckpoint(..., monitor = "loss", save_best_only = True)` So the best model is defined as the...

Ok, so I solved the issue with the `test_representations.py::test_topic_reduction_edge_cases` It just needed one line of code to `old_documents = topic_model._sort_mappings_by_frequency(old_documents)` Therefore, the test will actually check for ordered, and unchanged,...

> Awesome PR! Very clean and focused on the most important components here. I left a couple of minor questions that have mostly to do with variable naming. > >...

Ok I closed this PR because i found a much cleaner solution. I will make a new one, you will notice it has a minimal change that might not even...