Repetition Penalty
Noun · AI & Machine Learning
Definitions
A decoding adjustment that discourages a model from repeating the same tokens, words, or phrases too often during generation. Repetition penalties help reduce loops and dull redundancy in outputs.
In plain English: A setting that discourages the model from repeating itself too much.
Example: "They increased the repetition penalty after the assistant began echoing the same disclaimer in every paragraph."