How to calculate the f1_score in case of multilabel classification problem

Question

So I am doing a document classification and the class of the documents can be from [1,2,3,...,10].

Now I feed the model one document, I