Model download interrupted and turned into a corrupted model #461

Hi,

I was using the llama model for the first time, but my network stopped working so I killed the cli window while downloading it. After changing the network, I retried using the "flm run llama3.2:3b", but it didn't started downloading again. Instead, it launched the server using this faulty/partially downloaded model of llama, that runs but doesn't outputs an answer neither stop the request.

Running for the first time ("~/.config/flm/models" is empty):

After killing the download and running again:

If I delete the folder with the model, I am to run the download again, and after that, the model runs properly:

I'm currently using the release 0.9.37 for Ubuntu 24.04.

I will try solving it by myself, but its good to register it here. It doesn't look very complicated, maybe using ".partial" before finishing the download would solve it.

Thank you very much.