AI Campus • Practical resources
Machine Learning & Model Fine-Tuning
Separate model adaptation from quantization and compare changes using held-out examples.
Before you start
- Python, training/inference fundamentals and an understanding of training/validation splits. Training is optional and requires suitable hardware.
The reading and planning exercises need no paid course or API. Optional inference, training, IDE features and hosted services may require hardware, accounts or paid access. Use the local or manual alternative before spending money.
Start with the official learning material
Free-to-read references checked 3 November 2026. Reading documentation does not activate a provider or create a subscription.
Practice with a concrete exercise
1. Choose the intervention
Read the PEFT overview. Write a decision table comparing prompting, retrieval and parameter adaptation for a vocabulary task. Explain why quantization changes numerical representation and does not itself teach new domain facts.
2. Prepare a clean split
Author 20 toy input/output examples and reserve five for testing. Check for duplicates and near duplicates across splits. Record licensing and avoid sensitive real-world medical/legal records.
3. Evaluate within your hardware budget
Use the local-model project to measure inference first. If you later train an adapter using the official examples, record base model, adapter settings and held-out results. If training cannot run locally, deliver the dataset/evaluation plan and label training unperformed.
Check what you learned
- Held-out examples never enter the training data.
- The base model and adapted model use identical test cases.
- Measured speed/memory changes are distinguished from quality changes.
Keep your artifact, observed checks and remaining limitations in a short README. Mark any exercise you only designed rather than executed.