mediumAI Engineering

What is the difference between fine-tuning, LoRA and prompt engineering?

Q: What is the difference between fine-tuning, LoRA and prompt engineering?

**TL;DR.** Three levers, increasing cost: **prompt engineering** changes the input (no training), **LoRA** trains small adapter matrices, **full fine-tuning** updates all weights. Default path: prompt (+RAG for knowledge) first; fine-tune for **style, format, narrow-task specialization** — never to add facts. prompting edit the context LoRA train tiny adapters <rect class='d-box-muted' x='312' y='30' width='132' heigh

840 views

Understand the problem

Three ways to change model behaviour — weights, adapters or context — and how to pick.

fine-tuningloraprompt-engineeringadaptation

Attempt it yourself

Sketch your approach before reading the solution — that's what interviews test.

Stuck? AI Nudge Available

Get a conceptual hint to guide your logic without spoiling the final implementation.

Study the solution

The solution is waiting

Give it an honest attempt first — then compare your thinking with the full walkthrough.

Read the code

LoRA in one glance (PEFT-style)

from peft import LoraConfig, get_peft_model

config = LoraConfig(r=16, lora_alpha=32,
                    target_modules=["q_proj", "v_proj"],
                    task_type="CAUSAL_LM")
model = get_peft_model(base_model, config)
model.print_trainable_parameters()
# trainable params: 4.2M || all params: 7B || trainable%: 0.06

Join the discussion

Discussion (0)

No responses yet. Be the first to share what you think.

How do multimodal models handle images, audio and other non-text inputs?

How do you estimate and control LLM costs in production?