You now know the tools that make fine-tuning practical.
Takeaways:
- Hugging Face ecosystem is the foundation (Transformers, PEFT, TRL)
- Unsloth optimizes for consumer GPU training
- Axolotl provides configuration-driven reproducibility
- LLaMA Factory offers a beginner-friendly web interface
- DeepSpeed and Accelerate handle distributed training
Next, I'll show you how to deploy and evaluate your fine-tuned models.