Hello,
During the species creation in M, steps 1,2, and 3 out of 5 go well and after a few seconds at 4/5, the PC hard crashes. The screen goes black and starts a boot up again. It looks like the crash happens such that no error logs are created by the system and only after boot up does it recognize something went wrong.
Is it possible that the step that trains the denoising model could completely fill the GPU memory? I thought the training only uses two refined half-maps and not individual particles themselves. That's the only thing I can think of that might cause the crash.
Best,
Dennis