Include real word embeddings for TNTM, ETM

AnFreTh / STREAM

A versatile Python package engineered for seamless topic modeling, topic evaluation, and topic visualization. Ideal for text analysis, natural language processing (NLP), and research in the social sciences, STREAM simplifies the extraction, interpretation, and visualization of topics from large, complex datasets.

https://stream-topic.readthedocs.io

MIT License

33 stars 7 forks source link

Include real word embeddings for TNTM, ETM #62

Open AnFreTh opened 3 months ago

AnFreTh commented 3 months ago

move preprocess word embeddings to dataset class
make word embeddings attribute of dataset class
decouple from datamodule logic
Should be identical to sentence embeddings logic
use real word embeddings not sentence transformers (optional, maybe for later)