Generative AI & LLMs
Post-training
Training after pretraining that shapes a model's usefulness, safety, and instruction-following.
Example
After broad training, the model is adjusted to follow instructions and avoid unsafe responses.
Why it matters
It makes broadly learned abilities easier to use in practical products.
You'll hear it
“The newer version follows our formatting requests more consistently.”
Can you control it?
Set by the team
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Does post-training replace all earlier learning?
- No. It usually builds on the starting system rather than teaching everything again from nothing.
- Can it introduce new weaknesses?
- Yes. Improving one behavior can affect other abilities or create unwanted habits.
- Does it always use feedback from people?
- No. Different approaches can use human feedback, computer-generated checks or other learning signals.