AI conversations come with a lot of new words. Here is a broad collection of common AI and LLM terms, each explained in one or two sentences, with examples wherever useful. Jump to any definition using the table of contents beside this article on desktop or above it on smaller screens. Terms are grouped by topic; use your browser’s Find feature (Ctrl+F or Cmd+F) if you already know the word. ...
I have been running models with Ollama and vLLM. Then I stopped to ask: what is actually happening behind that API call? A learning chat turned into these notes. Let’s follow one request all the way through: Why did my Kubernetes pod restart? The container’s last termination reason is OOMKilled. This is an illustrative example, not a live diagnosis. We will use a hypothetical 2B model throughout. Part 1: The Mechanics 1. An answer grows one token at a time An LLM, or large language model, is a neural network trained on language. For the text-generating models here, the basic loop is: use the text so far to predict what comes next. ...