Skip to main content
ModelTerms

Comparison

LoRA vs QLoRA

LoRA and QLoRA are both common AI/LLM terms but cover different ideas. Here is a quick side-by-side.

When you would reach for LoRA

When full fine-tuning is too expensive or you want swappable specialized adapters.

Fine-tuning Llama-3-8B for a domain on a single A100 with LoRA.

When you would reach for QLoRA

When you want to fine-tune a frontier-sized open model on a single GPU.

Fine-tuning Llama-3-70B on a domain corpus on a single A100.

Frequently asked

What is the difference between LoRA and QLoRA?

LoRA: LoRA is a parameter-efficient fine-tuning method that freezes a model's original weights and learns small low-rank update matrices alongside them. Cheap fine-tuning on a single GPU. QLoRA: QLoRA fine-tunes a 4-bit quantized base model with LoRA adapters, letting you train 70B-class models on a single 48 GB GPU at near-full fine-tuning quality.

When should I use LoRA vs QLoRA?

When full fine-tuning is too expensive or you want swappable specialized adapters. When you want to fine-tune a frontier-sized open model on a single GPU.

Are LoRA and QLoRA the same thing?

No. LoRA is training; QLoRA is training. They are related but address different parts of the AI stack.