Fine-Tuning and Compression of Large Language Models
Workshop, Purdue University, Rosen Center for Advanced Computing, 2026
This intermediate-level workshop introduced practical methods for fine-tuning and compressing large language models. Topics included full fine-tuning versus parameter-efficient fine-tuning, Low-Rank Adaptation (LoRA), QLoRA, key LoRA hyperparameters, model quantization, and GGUF conversion for local deployment.
The workshop also provided a step-by-step walkthrough of fine-tuning Mistral 7B with Unsloth and running the resulting model locally with Ollama, with an emphasis on the accuracy, efficiency, and deployment tradeoffs among different fine-tuning approaches.