Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
Unstructured data visualization
Explore unstructured text data visualization with sentence transformers, UMAP, LLMs, and autoencoders. Experiment with embedding and summarization models on diverse datasets.
I’ve been building an open source tool for visualizing unstructured text data using sentence transformers, UMAP, LLMs and sparse autoencoders. The tool makes it convenient to experiment with various open models for both embedding text and summarizing clusters. I’ve been exploring several open datasets from fun things like 50k reddit dad jokes to more serious things like 100k samples of large training corpus or 500k pubmed journal abstracts.
Embeds, UMAP-reduces, HDBSCAN-clusters, and LLM-labels unstructured data for latent space visualization.
Latent Scope structures unstructured data using embeddings and LanceDB vector search.
Compose Email
Loading recent emails...