Glossary · Term

zero-shot

← all terms

Definition

Plain language

When a model is tested on a task with no examples and no extra training, just asked to do it cold.

As stated in the literature

An evaluation or usage setting in which a model performs a task without any task-specific examples or fine-tuning, relying entirely on what it learned during pretraining.

Also called: zero shot

Why it matters: It tests whether a model can handle new tasks straight out of the box, which matters because you can't always supply examples for everything you want it to do.

For example, you ask a model to translate a sentence into a language without showing it any sample translations first, and it simply attempts the task cold.

Heard on the show

“Empty library, zero skills: about 5% zero-shot success.”
Episode 194 — How a Robot Builds a Debugging Notebook It Can Read, Edit, and Hand to Another Robot

Mentioned in 8 episodes

  1. 194
    How a Robot Builds a Debugging Notebook It Can Read, Edit, and Hand to Another Robot
  2. 189
    Why Phone Agents Ace the Test and Crash on Your Actual Phone
  3. 091
    When Better Fine-Tuning Can't Help: A Geometric Impossibility in LLM Causal Reasoning
  4. 078
    Training a Markdown File: When LLM Self-Improvement Borrows the Discipline of Neural Net Training
  5. 064
    When Agent Memory Stops Being a Database and Starts Being a Skill
  6. 055
    Why LLM Judges Flip Their Verdicts When You Change the Question Format
  7. 051
    Why Parallel Sampling Plateaus, And What Evidence Graphs Do Instead
  8. 033
    Echo: The Paper Arguing You Never Needed a KV Cache for Retrieval

Related terms