Will the model give good results if it is trained to generate an abstractive summary of a large piece of text?

Will the model give good results if it is trained in generating an abstractive summary of 3-4 sentences?

If not what changes would you suggest for good results?