Practical guide: Fine-tuning Qwen3 with LoRA

​Ivan Potapov - Research Engineer, Zalando SE

​In this workshop, we’ll show how to fine-tune Qwen3 with LoRA using two powerful alignment techniques. With KL-anchored SFT, you’ll keep the base model’s quality while adding new styles and formats. With β-tuned DPO, you’ll control how strongly preference data steers the model, balancing domain focus with general ability.

​You’ll learn how to set up data, run training, evaluate with reward models and MT-Bench, and deploy LoRA adapters in production. By the end, you’ll know how to fine-tune Qwen3 for your tasks while avoiding drift and forgetting.

​

​About the speaker:

​Ivan Potapov is a Research Engineer at Zalando, specializing in search. He has taught workshops on data engineering, AI agents, and LLM alignment, helping practitioners bridge software engineering with applied machine learning.

​

​DataTalks.Club is the place to talk about data. Join our slack community!