Intermediate · Generative AI

PEFT (Parameter-Efficient Fine-Tuning)

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. A family of methods that adapt large pretrained models by training only a small fraction of parameters, drastically cutting compute and storage.

Technical Definition

A family of methods that adapt large pretrained models by training only a small fraction of parameters, drastically cutting compute and storage.

How it works

PEFT methods include LoRA (low-rank adapters), prefix tuning, prompt tuning, IA³, and adapters. The base model stays frozen while a small set of new parameters (often <1% of the original) is trained. Multiple task-specific PEFT modules can be swapped on top of one shared base, enabling cheap multi-tenant serving. Hugging Face's peft library is the standard implementation.

Related Concepts

• Fine-Tuning — Adapting a pre-trained model to a specific task by continuing training on a smaller, task-specific dataset.
• LoRA (Low-Rank Adaptation) — A parameter-efficient fine-tuning method injecting small trainable matrices into frozen pre-trained layers.
• QLoRA — Fine-tuning method that adds LoRA adapters on top of a 4-bit quantized base model, letting 65B LLMs be trained on a single 48 GB GPU.