Make a distributed trainer restartable by packaging step and weights into a serializable checkpoint without sharing mutable lists.
Required APIdef make_checkpoint(weights, step):
return {'step': ..., 'weights': [...]} Think through the mechanism first if you want the extra reasoning step. It never blocks the editor.
make_checkpoint([1.0, 2.0], 7)a restart needs both progress and parameters
make_checkpoint([], 0)checkpoint format is defined before model scale
2 hidden edge tests run after the visible contract passes.
Run Tests to see the contract verdicts here.