Custom tagging model generates blank pos_ output

[funghetto]

I've trained my model (for more info see my old issue.
On a little testing program:

import spacy

nlp = spacy.load("it_core_news_sm")
nlp2 = spacy.load(r"C:\Users\Michele\OneDrive\Documents\C\Programmaraffa2\Trainingmodel\UD_Italian-MarkIT-r2.10\Fine\model-last")

doc = nlp("Roma è la capitale d'Italia.")
doc2 = nlp2("Roma è la capitale d'Italia.")

for l in doc:
    print(l.pos_)
for k in doc2:
    print(k.pos_)

I'm getting this output:

PROPN
AUX
DET
NOUN
ADP
PROPN
PUNCT

Can someone give me any clue on why it is not tagging the phrase?

[adrianeboyd]

Check that your pipeline contains a morphologizer. Using the default spacy CoNLL-U converter, XPOS is saved as token.tag_ and UPOS is saved as token.pos_. token.tag_ is output by the tagger component and token.pos_ + token.morph are output by the morphologizer component.