If you like, you can use the top_k parameter to specify how many predictions to return:

from transformers import pipeline
mask_filler = pipeline("fill-mask", "username/my_awesome_eli5_mlm_model")
mask_filler(text, top_k=3)
[{'score': 0.5150994658470154,
  'token': 21300,
  'token_str': ' spiral',
  'sequence': 'The Milky Way is a spiral galaxy.