why do fibroblast,endothelial,epithelial exit in immune_ALL_High_model?

[echovv]

Hi,
Super helpful tools! Here is my question:
When I annotate the human spleen dataset using the immune_ALL_High_model, I get the figure below. I noticed that a portion of the fibroblast cells remains unassigned. I'm wondering what could be the reason for this. I might think is the different property of fibroblast, so I checked the relevant scientific paper, but it doesn't mention any stromal cells. So, where do the trained fibroblasts originate from, and what is the tissue source in this case?

Ziwei

[ChuanXu1]

@echovv, Fibroblasts and other non-immune cells act as umbrella categories to provide flexibility for annotating potential non-immune cells in the query data. Fibroblasts used in the training data are from thymus.