What can you do if the work planned for a sprint is completed before the end of the sprint?

Never mind if you've completed the work you planned to do. It was only ever a forecast. Have you actually met your Sprint Goal commitment?

If their commitment has been honoured for that Sprint, the Developers can throw a party, go on holiday, or do whatever they want to do. Perhaps they might go on a training course, or take the opportunity to remediate some technical debt. In the Sprint Retrospective, they'd consider adjusting their capacity calculations for the next Sprint Planning session. The Sprint will end when the time-box for it ends.

Sprints have a set timebox for a reason. It helps to create a cadence that creates consistency.  This helps with predictability.  If you change that cadence, you start to mess with other factors.  @Ian is spot on with his answer.  There is nothing wrong with finishing work early, but the real gauge is whether the Sprint Goal has been satisfied and the value that was planned is delivered.  If that is true, then the team finds itself in a great situation to do some work on technical debt, research some new technologies, spend some time refining the Product Backlog, have a quick "hack-a-thon" and build something fun just to do it. All of those things can provide value to the team.  Also the non-product related activities could be good for morale.  Celebrate the accomplishment, take a little time to do something less stressful, and then discuss this in the retrospective to understand why and how it occurred.  It could be a fluke or it could be that the team is getting more efficient and can try to do more in each Sprint.  Let the team decide.