Currently: -134.9478/(-306.8952)

Should be: -306.8952 / -134.9478

Yes, thank you for the correction. We are now revising the text of that section to:

"The spam model assigns a log likelihood of -134.9478 which is much better than the null model's -306.8952."