Boltz fails at structure prediction when running locally, but running through YASARA seems to work fine

[ikor93352]

Running boltz directly seems to give completely incorrect results regarding structure and affinity. Using a template doesn't help. I'm able to use boltz in YASARA and get more accurate looking results so I'm not sure what the issue is.

The affinity example protein folded through boltz in YASARA Same structure folded through boltz directly