A numerical representation of a piece of information
This week one of my delights was being exposed to the world of embeddings, which is a method commonly used in the machine learning space when wanting to capture the semantic meaning of what is being embedded.
It is another example of the prominence of vectors in this space and how what is ultimately a simple mechanism, can provide great value (when applied appropriately). The linked blog post does a good job of running one through a query -> FAQ scoring exercise. I did worked through that using a personal data set and it was fun. Recommend!
As an aside, I am super impressed with what Hugging Face is doing to democratize machine learning stuff. The field has felt long-inaccessible, but here is a space with the right community, tooling, and guidance to open the door to many more.
I started a side project and chose fly.io for my deployment. I was impressed with the quick startup time. I'll be attempting a project which blends a bit of machine learning and live lookups at the edge to provide a snappy result. More on what that actually means soon; hoping to sprint on a 2-week MVP when things calm down with life / work. (do they ever? ... yes)
- Nope I am seeing Jordan Peele's newest flick later today. I appreciate his work because he doesn't hold back, isn't afraid of some in-your-face social commentary, and is a great writer overall.