AI engineering glossary · Prompt context engineering
What is logit bias?
Logit bias: A parameter that adds a fixed number to the score of chosen tokens before one is picked, making them more or less likely, or effectively banned. It can force a choice among labels or suppress unwanted words.
Logit bias: an example
Give the tokens for 'Yes' and 'No' a large positive bias so the model answers with only one of them.
Related terms
More Prompt context engineering terms
- Logprobs
- Long-context prompting
- Lost in the middle
- Maieutic prompting
- Majority label bias
- Map-reduce summarisation
- Markdown formatting in prompts
- Max tokens
Where you learn it
Logit bias is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.