Logits are the raw numerical scores that language models compute for each token in their vocabulary before making a selection. These scores are converted to probabilities and used internally for token selection. Accessing and analyzing them directly opens up possibilities for controlling and understanding model behavior.

We'll cover common sampling techniques like temperature adjustment, top-k, and top-p filtering, along with more advanced approaches like beam search. The talk introduces practical applications through custom samplers.

Logits also provide insights into model uncertainty, help detect potential hallucinations, and enable fine-grained control over generation.

Throughout the session, we'll run live experiments demonstrating these concepts in action. We'll explore helpful libraries and tools. You can get practical insight which you can apply to your own projects. No advanced math or deep ML expertise required—just a curiosity about how language models make decisions.