Question $2.$ N$-$gram language models $[5$ Marks$]$

"Erring is human. To err is human. Seeing is believing. To see is to believe."

Given above training data, use bigram language model to compare the likelihood of below test sentences. Note: To handle zero probability, use interpolation with weights $0\mathrm{.}3$&$0\mathrm{.}7$ for unigram and bigram respectively. Don't account for punctuation ie$.,"."$ for the modelling and predictions.

a$)$ Test Sentence $1$: Seeing is human

b$)$ Test Sentence $2$: To believe is human