I would to like to use an Electra model instead of a Bert model.
In order to do this, do I need to replace only the path to the model in my code?
So, for example, instead of “bert-base-uncased” I have to use " electra-base-uncased-discriminator".
Hi Sergio, there is an example in Electra manual that you can follow : https://huggingface.co/transformers/model_doc/electra.html#tfelectraforsequenceclassification
Thank you, Jung. From the doc it seems I need to change only the bert model with the electra discriminator. I’ll do some tests