# unsupervised (blogs) — RSS Amplifier Recent posts from the 1 feeds in the RSS Amplifier directory that cover unsupervised. Page: Feed: --- ## [How you can run the same Python distribution on EKS and AWS Lambda (using UV)](https://yonatanlou.github.io/blog/how-you-can-run-the-same-python-distribution-on-eks-and-aws-lambda/how_you_can_run_the_same_python_distribution_on_eks_and_aws_lambda/) _2025-11-08 · Yonatan Lourie_ I recently tackled a problem that ate up a few hours of my day, and I'm documenting the solution for anyone who hits this same wall in the future. The Setup I'm doing my development with uv , and I was working on a microservice that serves an ML model via API. It's a FastAPI application running at on an EKS cluster in Kubernetes. Everything was working great. Then came another requirement: we… ## [Things I learned in my research roles over the last 5 years](https://yonatanlou.github.io/blog/things-that-i-learned-in-my-latest-research-roles/) _2025-09-22 · Yonatan Lourie_ In the last ~5 years I've done a lot of research across various topics and places. I learned a ton, so this is where I'm dumping the main lessons from those years. \[Statistics Student, Analyst in the public sector, Analyst + Data Scientist at Forter, MSc in deep learning, LLM research in academia\] This list is random, not sorted by importance 1. Always write clean code Most of the time, you're… ## [Unsupervised text clustering using GNN (Graph Auto Encoder)](https://yonatanlou.github.io/blog/unsupervised-text-clustering-gnn/unsupervised-gae-clustering/) _2025-06-18 · Yonatan Lourie_ Graph neural networks (GNNs) have received a fair amount of attention over the past few years. That said, some of the initial excitement has faded-especially in certain research domains. Part of this decline is due to the rise of transformer models, which in many ways behave like fully connected GNNs. This has led some people to question whether GNNs are still relevant or necessary. ( Transformers… ## [Evaluating Hierarchical Clustering Beyond the Leaves 🌳](https://yonatanlou.github.io/blog/Evaluating-Hierarchical-Clustering/hierarchical-clustering-eval/) _2025-01-12 · Yonatan Lourie_ When working on hierarchical clustering tasks, evaluation is often limited to the leaves of the tree or relies on plots that look impressive but don't tell the whole story.What if we could assess the entire tree structure? Wouldn't that give us a more nuanced view of our clustering methods, linkage techniques, or embedding models? In this short article, I'll introduce you to a clever method… ## [Jaccard index for unsupervised clustering](https://yonatanlou.github.io/blog/jaccard-index-unsupervised/Jaccard_index_for_unsupervised_clustering/) _2024-11-10 · Yonatan Lourie_ TL;DR - Don't use sklearn jaccard\_score function when dealing with unsupervised clustering. The Jaccard Index is a similarity measure used to compare the similarity and diversity of sample sets. It's calculated as the intersection divided by the union of sample sets. While it's a useful metric, its direct application in scikit-learn's jaccard\_score is not suitable for unsupervised clustering… ## [How to use the Qumran scrolls for NLP tasks using text-fabric](https://yonatanlou.github.io/blog/data-pipeline-for-qumran-scrolls/data-pipeline-for-the-qumran-scrolls-using-text-fabric/) _2024-10-30 · Yonatan Lourie_ When I first started my thesis, I inherited the Dead Sea Scrolls project from another student in my lab. He provided a large file that looked something like this: I also inherited a complex spaghetti code that parsed this data into readable Hebrew text. These files (from the image above), were created by Martin Abegg and contain the raw text of the Dead Sea Scrolls, with morphological data on each… ## [Moving from Jekyll to Eleventy.](https://yonatanlou.github.io/blog/firstpost/) _2024-10-29 · Yonatan Lourie_ I got tired of the Ruby environment. I wanted to start using javascript. That's all