Table of Contents

Fine-Tuning Gemma 4 with QLoRA for Tool-Aware Support Agents

From Customer Support Chatbot to Tool-Calling AI Agent

Synthesizing Tool-Calling Training Data

Why Use Synthetic Tool-Call Trajectories?

What We Will Build

Configuring Your Development Environment

Setup and Imports

Defining the Shared Configuration

Configuring QLoRA

Loading the Bitext Customer Support Dataset

Preparing the Agentic Training Dataset

Reloading a Fresh Gemma 4 Base Model

Training the Agentic Model

Saving the Fine-Tuned Gemma 4 LoRA Adapter

Performing a Quick Inference Check

Merging the LoRA Adapter and Publishing to the Hugging Face Hub

Summary

In the previous lesson, we fine-tuned Gemma 4 E2B-IT for customer support using Supervised Fine-Tuning (SFT) and QLoRA (Quantized Low-Rank Adapter; Dettmers, 2023) .…