"Floating point exception" when run inference_recon.py

[hanxunyu]

Has anyone encountered this issue?
It appears that the error occurs in past_key_values = self.forward_cache_update_text(past_key_values, **generation_input).
Any insights on how to fix this would be greatly appreciated.