bert_document_classification
bert_document_classification copied to clipboard
Does the model only do multilabel predictions?
I am using a custom dataset for a multiclass classification problem where my data points only belong to one class (mutually exclusive case, only 1 label per sample and not many like in multilabel), does the model implicitly understand that the problem is multiclass classification based on the training data labels? If not, how does one force the model to only output one label per sample?