I have conducted a session on “Understanding AI: Deep dive behind the curtain” in our office Cheq, a Cantaloupe Platform and it was quite the experience. I have used AI and LLM interchangeably here.

I have been in front of TV Cameras, Debates, On Stage in front of a large crowd and the exhilaration that comes from talking to people in that kinda setup, always fires me up.

In this session, i discussed among other things, Tokenization, Embedding, Weight based decision making, Inner workings of the transformer, Model context and context window, How next token prediction works, and how Post processing is done for generated responses.

There was quite a few good questions from the QA session, and it was amazing to see the interest my team mates have to understand AI better. I am always of the opinion that to use a tool, it needs to be understood properly.

If you understand how AI actually works, you can make it do anything you want, efficiently and exceptionally well. Which in turns reduces cost, failed responses and in the long run, improves your peace of mind.

Here to hoping to conduct more sessions, soon.