Large Language Models (LLMs) have transformed the way artificial intelligence interacts with human language. While these systems are immensely powerful, their inner workings can feel like a mystery to most.
This article aims to simplify the complexity behind LLMs, breaking down advanced concepts such as neural networks, and transformers in a way that's easy to grasp.
Step 1. Indiscriminately hoover up any text you can find in the name of training data.
Step 2. ???
Step 3. Profit.
That sounds about right! but Step 3 seems working well for them...
I'm not so sure about that.
Companies are burning through huge amounts of money to train and run these models,
but the revenue isn't keeping up.
Billions of investment, but where's the revenue?
https://www.axios.com/2024/07/12/ai-bubble-revenue-missing
Yeah, Nvidia's pocketing the revenue, but it seems we’re in an era where revenue and profit do not matter, money goes where the hype is...