How to save the trained model for future training?

[Martisum]

I hope to improve the training beyond 15000 steps, but each improvement requires starting from scratch, which is very time-consuming. I hope to use the command-line parameters provided by train.py to save the model, but I have been failing.
The parameters I used are as follows
--start_checkpoint D:\Work\AI\PriorDepth-GaussianSplatting\output\eval_gsopt4 --save_iterations 15000 30000
But the model cannot correctly read the training progress I saved. And I found that it seems that the model only saves ply files. Does this mean that the source code cannot achieve the functionality I mentioned?