Capstone: Fine-Tune a Model on Your Own Data

You've covered the full pipeline from theory to deployment. Now apply it. This capstone presents three real challenges to test your skills — then you earn your certificate.

What you covered in this course

Module 1: Why Fine-Tune

When prompts fail. The spectrum from prompting to pre-training. When fine-tuning wins and loses.

Module 2: Dataset Preparation

The three formats. Five quality filters. JSONL output for training frameworks.

Module 3: LoRA and QLoRA

Low-rank adaptation. Rank and alpha. QLoRA on consumer hardware.

Module 4: Instruction Tuning and RLHF

SFT, reward models, PPO loop. DPO as the simpler modern alternative.

Module 5: Evaluation

Four-layer eval stack. Catastrophic forgetting detection. LLM-as-judge.

Module 6: Deployment

Merge LoRA, quantize with GPTQ or GGUF, serve via FastAPI or HF Endpoints.

Your capstone challenges

Challenge 1 — Core

Build a fine-tuning dataset from scratch

Choose a domain you know: your industry, your job, or a topic you're deeply familiar with. Collect or write 100 instruction-response pairs. Apply all five quality filters from Module 2. Format them as JSONL. Write a short paragraph explaining: what task does this dataset teach, and what format consistency rule did you enforce?

Challenge 2 — Technical

Configure a LoRA fine-tuning run

Using the code from Module 3, configure a LoRA fine-tuning run for your dataset. You don't need to actually run it (that requires GPU resources). Instead: write out the LoraConfig with your chosen rank and alpha, explain why you chose those values, and list which target modules you would use and why. Then write the evaluation plan from Module 5: what metrics would you use, and how would you check for catastrophic forgetting?

Challenge 3 — Strategic

Make the deployment decision

You've fine-tuned a 7B model for internal customer support at a company handling 300 support tickets per day. Average response length is 150 tokens. The team needs responses in under 2 seconds. Internal security policy does not allow data to leave the company's servers. Write a deployment plan: Which quantization format? Which serving approach? Why? What would the estimated infrastructure look like?

Certificate of completion

Enter your name below. The certificate generates automatically. Click to edit any field.

AJ University — Certificate of Completion
This certifies that
Your Name
has completed
Fine-Tuning Large Language Models
Arjun Jaggi
arjunjaggi.com

"Just completed the Fine-Tuning Large Language Models course for free at arjunjaggi.com — covered LoRA, QLoRA, RLHF, evaluation, and deployment end to end."

What to learn next

AI Agents

Deploy fine-tuned models as domain-specific agents that use tools and memory.

Prompt Engineering

The fastest way to improve a model without fine-tuning. Use both techniques together.

How LLMs Work

Deep-dive on transformers, attention, and scaling laws that underpin everything you just learned.

Want to apply this at your company?

Book a free 30-minute call to discuss your fine-tuning use case and what it would take to get a pilot running.

Book with Arjun