Questions about classification models

Models
1

Hello,

I’m new to HuggingFace and I’m struggling to understand some details of the classification models (TFAutoModelForSequenceClassification):

  • In all snippets given in the documentation, the target is never specified explicitly. Does the model expect it to be named as “label”, “labels” or something? I haven’t found this specification in the models’ documentation.
  • The tokenizer produces 3 fields: input_ids, token_type_ids and attention_mask. Does the model expect them and is it able to handle them properly? Once again, I haven’t found anything about it in the documentation.

Thanks in advance,
Mark

1 Like