Glossary · Term

logits

← all terms

Definition

Plain language

The raw numbers a model produces before they get turned into probabilities over words.

As stated in the literature

The unnormalized scores output by a model's final layer, converted to a probability distribution via softmax.

Also called: logit

Why it matters: Operating on logits rather than probabilities — for sampling, calibration, or distillation — is standard practice and avoids precision issues.

For example, a model might output logits of (2.1, 0.4, -1.0) over three tokens, which the softmax then turns into probabilities of roughly (0.78, 0.14, 0.08).

Heard on the show

“For the frontier models, where they can't read logits and have to sample outcomes, candidates get screened on one batch and confirmed on another.”
Episode 243 — How a Hundred Meaningless Word Choices Add Up to Flip a Model's Answer

Mentioned in 7 episodes

  1. 243
    How a Hundred Meaningless Word Choices Add Up to Flip a Model's Answer
  2. 230
    Why AI Survey Panels Break Before the Dice Ever Roll
  3. 094
    Chain-of-Thought Monitoring Fails Across Languages, and Worst Where It's Needed Most
  4. 084
    Terminal Agents Get Free Supervision From The Tokens We've Been Throwing Away
  5. 073
    When Three LLMs Talk to Each Other, Their Ideas Quietly Stop Moving
  6. 037
    Why Hallucination Detectors Miss Stale Facts: A Geometric Story About What Models Know But Don't Say
  7. 018
    Language Models Compute the Rational Move, Then Override It

Related concepts

Related terms