ValueError: cannot compute similarity with no input about lbl2vec HOT 3 CLOSED

The keywords 'Crack', 'Broken' and 'Breakage' were not learned by the model and therefore unknown to it. Probably those were all keywords for your class but can't be used to compute a label vector because they are unknown. This results in an error.

This could have different reasons. The simplest explanation is that you used capitalized keywords, but the model only knows words that are lowercase. In this case, just convert your keywords to lowercase.

Another explanation could be that the keywords don't appear in your training corpus or have a low frequency. In this case I suggest you try some different keywords or add some more training data that the model can learn those keywords.

from lbl2vec.

The unknown keywords are already skipped by default for computing the label vector. But when all keywords are unknown to the model, no keywords are left for label computation. This probably resulted in the error.

from lbl2vec.

Is it possible to skip those terms that aren't in the document?

from lbl2vec.