why the predicted cell types are sometimes incomplete

[zhujuntong]

I am sorry that there may be some problems with my expression before. I would like to ask why the predicted cell types are sometimes incomplete and the predicted results are inconsistent every time.

[QingJun3]

I'm having a similar problem. The provided code gives different results every time after running it on the TonsilBE dataset. I looked closely at the code and found that SEED was not set, but even if it was, the results were inconsistent, most of the time only about 60% accuracy.