Learn Before
Short Answer

Analyzing the Mechanism of Prefix Tuning

A parameter-efficient fine-tuning technique involves keeping a large language model's parameters frozen while introducing a small set of new, trainable vectors. These vectors are prepended to the sequence of hidden states at each layer of the model. Analyze the primary reason why these vectors are added at every layer, rather than only at the initial input layer, to steer the model's behavior for a new task.

0

1

Updated 2025-10-06

Contributors are:

Who are from:

Tags

Ch.3 Prompting - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences

Analysis in Bloom's Taxonomy

Cognitive Psychology

Psychology

Social Science

Empirical Science

Science