Checkpoint Manager RAM leakage

[skelsan]

What happened?

Hello,

I am experiencing a strange RAM leakage problem in Checkpoint Manager when I sort models and LoRAs into different directories.
If I do this manually via File Explorer and then click “Refresh” button in the program it takes 250-750 MB RAM.
If I use the “Drag and Drop” feature built into the program, it's even worse, every object (Model/LoRA) I move causes “Refresh” to be triggered automatically, consuming about 425 MB per object. For example moving 16 LoRAs to another directory resulted in a loss of 7 GB.
The expected behavior is to release RAM after the operation completes.
I can't say exactly when it started, but for the last two updates the problem is definitely present.
I've noticed that if I wait long enough and don't do anything in the program, the memory will be freed, but it takes a long time. To give you an example, 20 models were moved, which took up about 9.1 GB of RAM, after 30 minutes 1.1 GB was freed, after another 30 minutes 2 GB, then after 15 minutes 4.5 GB and after 10 minutes 1.5 GB. Then after 10 minutes, a strange thing happens and about 6 GB become taken again.
Currently to get around this problem I have to use file explorer and keep track of how many times I click on the “Refresh” button on the Checkpoint Manager tab.

Steps to reproduce

1. Open StabilityMatrix.
2. Click the Checkpoint Manager tab.
3. Select an object (Model/LoRA) and Drag and Drop it to another directory. Wait and see how the program eats RAM.
   OR
4. Select and move the object to another directory using explorer, then click the “Refresh” button on the Checkpoint Manager tab.
5. I expect to see RAM freeing a short time after the move operation is complete, but it takes a very long time.

Relevant logs

Version

v2.13.4

What Operating System are you using?

Windows