JavaScript implementation of UMAP

A 3D visualisation of clustered data using UMAP and three.js

an experiment using vector embeddings and t-SNE/UMAP to display embeddings in 3D using all-MiniLM-L6-v2 and transformers.js

tfjs + umap-js + USE

Luxury product intelligence platform, semantic search & auto-clustering over 29k fashion SKUs. Stack: OpenAI embeddings · UMAP · HDBSCAN · FastAPI · BigQuery · Next.js 16 · Cloud Run

Simple CLI to make plots of JSON Lines files

No description available

No description available

No description available

Interact with the Adobe User Management API from node.js

No description available

Simple d3 comparison of dimension reduction algorithms

SEO Clustering - Semantic keyword clustering with Spanish embeddings, UMAP visualization and D3.js

Full-featured document search system with semantic/keyword hybrid search, PII detection, multi-collection management, and interactive UMAP clustering. Qdrant + Ollama + Vue.js.

I evaluated the embedding models using Transformers.js within the plugin environment to benchmark real-world performance and scalability(Also tested UMAP and K-Means clustering).

An Interactive Single‑Cell Explorer built with Next.js Server Actions for real‑time UMAP visualization and ONNX‑powered DNN cell‑type annotation of free public scRNA‑seq datasets

An in-browser, serverless demo showing how large language models map text into semantic space. Sentences are embedded with Transformers.js and projected with UMAP into 3D; each point lights up inside a stylized brain volume to illustrate how meaning clusters in vector space.

I have made 2 visualisations: 1.⁠ ⁠Applied t-SNE to reduce dimensions and plotted it using three.js on 2D scatter plot. (As done in the Linkedin post) 2.⁠ ⁠⁠Create CLIP Embeddings using clip-ViT-L-14. Apply UMAP to reduce embedding dimensions to 2D. Plotted it on 2D scatter plot