Post

I spent 2 months learning about quantization and am extremely proud of the post I've written about it. I think these are some of the nicest visuals I've ever made, and I love how this compression technique invented in 1898 is being used on the bleeding edge in 2026. ngrok.com/blog/quantiz...

Lil' gallery of some of the visuals to entice you to click and justify my continued employment. The post starts by introducing what models actually _are_ under the hood, and what leads them to become so large they can't fit in your laptop's RAM.

Then I talk about how computers represent floating point numbers, the trade-offs involved, and how LLMs kinda don't care that much about precision (but not in that way.)

As part of the work on this post, I opened up 6 popular open models and analysed their parameters. This is a histogram that shows that model parameter values cluster close to 0, with the vast majority of values being very small.

There were a tonne of fun ways to visualise the precision of floats as you use fewer bits to represent them. This one shows the round-to-nearest behaviour between 16-, 8-, and 4-bit floats.

I also had a lot of fun learning the different ways that LLM degradation is measured after they've been compressed to a quarter or even an eighth of their original size. One of these measures, KL divergence, is all about how well 2 probability distributions overlap.

Beautiful work.

It's great! I was planning to start writing a series on quantization for my blog but this is so much better 😅 Now I don't need to write anything, I can just link to this.