Concept
Almost nobody needs to train a model.
Nearly every real use case is prompting, retrieval, or at most fine-tuning an existing model.
Concept
Training, Fine-Tuning & Inference
Understand the difference between training, fine-tuning and inference, and which one you actually need. Free, one sitting.
Start Quick LessonFree to explore · ~3 minutes
Why this matters
Someone says "we should train a model" and means five different things depending on the room. Training from scratch costs millions. Fine-tuning costs hundreds. Inference is what you pay per request forever. Getting these straight changes what you plan, budget and promise.
What you'll cover
Three Different Things
- Why the words get confused
- The three stages in order
- Relative cost
- Who does which
- Getting the vocabulary right
Pretraining
- What pretraining produces
- The scale involved
- Data and its problems
- Post-training
- Why you will not do this
Fine-Tuning
- What fine-tuning changes
- The knowledge misconception
- Data requirements
- Cost and effort
- Maintaining a fine-tune
Inference
- The cost that repeats forever
- What happens per request
- Levers that reduce it
- Serving considerations
- Forecasting inference spend
Choosing What You Need
- Start with prompting
- Add retrieval for knowledge
- Fine-tune for consistent form
- A decision tree
- Explaining the choice
What you'll understand
- Distinguish the three clearly in conversation
- Estimate the relative cost of each
- Say when fine-tuning is genuinely warranted
- Stop promising training when you mean prompting
A look inside
Three moments from this Quick Lesson
The real thing — not a mockup of it.
How it fits
Pretrain (once, huge) → Fine-tune (sometimes) → Infer (constantly)
Only the last one is a cost you carry on every single request.
Apply
Which cost repeats forever?
Pretraining
Inference
Fine-tuning
How it works
01
Understand the concept
A plain-language walkthrough of the idea itself, no prior context assumed.
02
See it in practice
A simple diagram or example showing how it actually fits together.
03
Apply what you learned
One quick check that you can recognise it, not just recall it.
Useful for
Product managersFoundersEngineering managersDevelopers
Ready to understand Training, Fine-Tuning & Inference?
Want to go deeper? Explore LLM Engineer →
Frequently asked
When is fine-tuning worth it?+
When you need a consistent format, tone or narrow behaviour that prompting keeps missing — and you have hundreds of good examples.
What will I learn?+
What each of the three actually involves, their relative costs, and how to pick the right one for a given goal.
Is this Quick Lesson free?+
Yes. Quick Lessons are free, short, and do not require a paid plan — sign in only to save your progress.