As mentioned above, I would like to continue to pre train 'xlm-roberta-base' with my own corpus, and then use it as an encoder. Is that okay? By the way, is the 'xlm-roberta-base' model used in the project original?
In addition, if I only use one of the languages, such as English, can I choose to use the single language model -- 'roberta-base'?
As mentioned above, I would like to continue to pre train 'xlm-roberta-base' with my own corpus, and then use it as an encoder. Is that okay? By the way, is the 'xlm-roberta-base' model used in the project original?
In addition, if I only use one of the languages, such as English, can I choose to use the single language model -- 'roberta-base'?
Thank you! A good job.