LLM (Büyük Dil Modeli) Fine-Tuning Uzmanlığı: GRPO, SFT, DPO
Fine-tune LLMs with DeepSeek! Learn SFT, LoRA, DPO & GRPO for practical reinforcement learning. Master advanced LLM training techniques.
Course Content
LLM (Büyük Dil Modeli) Fine-Tuning Uzmanlığı: GRPO, SFT, DPO is an all-levels Udemy course taught by Çağatay Demirbaş. It runs 3h 50m, is taught in Turkish, and includes a certificate of completion. Learners rate it 5.0 out of 5 from 11 reviews, with 51 students enrolled. It is a paid course, last updated April 2025.
Mastering Large Language Models (LLMs): From Fine-Tuning to Group Optimization
Dive into the world of Large Language Models (LLMs) and gain end-to-end expertise in both fundamental and advanced optimization techniques. This course provides practical, hands-on experience with key methods for building and refining powerful AI models.
We begin with Supervised Fine-Tuning (SFT), covering data preparation, and creating customized datasets using tokenizers and data collators. You'll learn the secrets to making large models lighter and more efficient using LoRA (Low-Rank Adaptation) and quantization, and discover how to integrate these techniques into your projects step-by-step.
Building on your SFT foundation, we'll explore Direct Preference Optimization (DPO). DPO allows you to directly incorporate user feedback, leading to models that are highly aligned with user needs. You'll learn how to format data for DPO, design effective reward mechanisms, and share your trained models on platforms like Hugging Face. We'll also delve deeper into the role of data collators in DPO, covering practical dataset preparation and transformation techniques for various scenarios.
A core focus of the course is Group Relative Policy Optimization (GRPO), a rapidly evolving technique delivering impressive results. Learn how to optimize model behavior not just for individual users, but also across communities or different user groups. This enables more systematic and effective adaptation of LLMs to diverse audiences and purposes. We'll cover the fundamental principles of GRPO and solidify your understanding with practical, real-world examples.
Throughout the training, we'll integrate LoRA, quantization, SFT, DPO, and especially GRPO with project-focused applications. By the end of this course, you'll confidently manage the entire LLM lifecycle – from data preparation to model fine-tuning and group-based policy optimization. Develop modern, competitive LLM solutions that prioritize both performance and user satisfaction with ease!
Want AI in your product? Hire the team behind TutorialSearch.io
LLM features, RAG, agents and automation wired into a real product — built by the devs who shipped this site.