How to guess, with spaCy, the frequency of a lemma or a word form in general language use?

I implemented with spaCy a toolkit for Text Analysis, structured in 2 levels:

• A. top level implements higher level functions plus interactive web views in Django;
• B. bottom level, linked directly to the spaCy library, provides, through web API, basic functions extending the spaCy functionality; it also uses spaCy corpora (DocBin) to persist
  reusable results of long-lasting or costly operations, such as the access to open resources on the web.

One extension in B is the annotation of token spans with references to nodes in lexical-semantic networks on the cloud, such as BabelNet (WikiData + WordNet + ...). In order to minimize the calls to the BabelNet web API, I would like to filter out lemmas and word forms very common in the general use of the language.

Question:

• Are there, in spaCy, methods to retrieve the frequency of a word type or of a lemma in the training data, considered as representative of texts produced in general language use?
• Do the spaCy vocabulary and the associated vectors keep some information related to said frequency?

In spaCy, there are indirect methods to retrieve frequency-related information for a word form or lemma based on the data the model was trained on, though spaCy doesn’t provide explicit frequency counts.

• SpaCy’s prob attribute in Lexeme gives a log probability based on the training corpus, which can be used as an indicator of general frequency.
• Vector norms can act as a rough frequency indicator but are less precise.
• For exact frequencies, consider combining spaCy with external frequency resources like wordfreq.