Implementation:Huggingface Peft FourierFTModel

Knowledge Sources	Huggingface_Peft
Domains	Deep_Learning, Parameter_Efficient_Finetuning
Last Updated	2026-02-07 14:00 GMT

Overview

Concrete tool for applying FourierFT (Fourier Fine-Tuning) to pretrained transformer models, provided by the Huggingface PEFT library.

Description

FourierFTModel is a tuner class that creates a FourierFT model from a pretrained transformers model. It adapts target modules by learning updates in the Fourier frequency domain rather than directly in the weight space, supporting both torch.nn.Linear and Conv1D layers. The method is described in detail in https://huggingface.co/papers/2405.03003.

Usage

FourierFTModel is typically created internally by calling get_peft_model with a FourierFTConfig. It can also be instantiated directly by passing a base model, a FourierFTConfig, and an adapter name.

Code Reference

Source Location

• Repository: Huggingface_Peft
• File: src/peft/tuners/fourierft/model.py
• Lines: 31-129

Signature

class FourierFTModel(BaseTuner):
    prefix: str = "fourierft_"
    # Inherits __init__ from BaseTuner:
    # def __init__(self, model, config, adapter_name):
    #     ...

Import

from peft.tuners.fourierft import FourierFTModel

I/O Contract

Inputs

Name	Type	Required	Description
model	nn.Module	Yes	The pretrained model to adapt
config	FourierFTConfig	Yes	Configuration for the FourierFT adapter (n_frequency, scaling, random_loc_seed, etc.)
adapter_name	str	Yes	Name identifier for the adapter, defaults to "default"

Outputs

Name	Type	Description
adapted_model	FourierFTModel	Model with FourierFT adapter layers injected into target modules

Usage Examples

Basic Usage

from peft import get_peft_model, FourierFTConfig
from transformers import AutoModelForCausalLM

model = AutoModelForCausalLM.from_pretrained("model-name")
config = FourierFTConfig(
    n_frequency=1000,
    target_modules=["q_proj", "v_proj"],
)
model = get_peft_model(model, config)

Related Pages

• Environment:Huggingface_Peft_Python_Core_Dependencies