amelius parent
Just ask an LLM to remove the personality from the training data. Then train a new LLM on that.
It will work, but at the scale needed for pretraining you are bound to have many quality issues that will destroy your student model, so your data cleaning process better be very capable.
One way to think of it is that any little bias or undesirable path in your teacher model will be amplified in the resulting data and is likely to become over represented in the student model.