Open-source Contributions

A collection of my open-source contributions, including minor fixes and improvements to existing projects. While I'm yet to make substantial contributions, I've started taking my first steps toward contributing to open-source projects.

FASTopic · June 2025

FASTopic is an efficient topic modeling method, introduced in NeurIPS 2024, that uses optimal transport and pretrained Transformer embeddings to discover topics and document-topic distributions.

Contributions:

  • Added a convergence monitoring feature that plots loss against epochs using model.loss_arr to track epoch-wise loss. [ PRs: #13 ]

BERTopic · Dec 2022

BERTopic is a topic modeling technique that leverages transformers and c-TF-IDF to create dense clusters allowing for easily interpretable topics whilst keeping important words in the topic descriptions.

Contributions:

  • Added update_topics() function with diversity and top_n_words parameters to customize topic representations without rerunning topic formation. [ PRs: #887, #888 ]