Jan. 16, 2024, 4:02 p.m. | Rodrigo Agundez

Towards AI - Medium pub.towardsai.net

Last Feature of ChunkDot

ChunkDot 0.3.0 now supports operations between 2 different input item embeddings.

Photo by Emile Guillemot on Unsplash

In my last 2 blog posts, I introduced ChunkDot, a multi-threaded matrix multiplication, and cosine similarity calculations at scale for dense and sparse matrices. I showed how it can calculate pairwise cosine similarity calculations for very big matrices containing either dense or sparse embeddings.

big blog cosine-similarity data science embeddings feature machine learning matrix matrix multiplication nlp operations photo python scale

Lead Developer (AI)

@ Cere Network | San Francisco, US

Research Engineer

@ Allora Labs | Remote

Ecosystem Manager

@ Allora Labs | Remote

Founding AI Engineer, Agents

@ Occam AI | New York

AI Engineer Intern, Agents

@ Occam AI | US

AI Research Scientist

@ Vara | Berlin, Germany and Remote