Model Training & Adaptation
Checkpoint
A saved snapshot of model parameters at a particular point in training.
Example
The team saves model weights midway through training so work can resume after an interruption.
Why it matters
It reduces lost work after interruptions and supports comparisons between stages of learning.
You'll hear it
“Resume from the last saved point rather than starting again.”
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 every checkpoint include the same information?
- No. Some contain only learned values, while others include additional training state.
- Is the latest checkpoint necessarily the best one?
- No. Later training can worsen some results or introduce overfitting.
- Can loading it reproduce the exact run everywhere?
- Not necessarily. Software, equipment and missing setup details can affect continuation.