QUESTION $4$

Originally BERT is defined in two sizes: BERT$-$base and BERT$-$large.

$(6)$ True

O False

QUESTION $5$

For a NLP model SuperGLUE is more difficult than GLUE.

$6$ True

False

QUESTION $6$

BERT is the acronym for Bidirectional Encoder Representations from Transformers.

True

O False

QUESTION $7$

Fine tuning $($step$2)$ in a BERT model takes more time than pretraining $($step $1).$

True

$6$ False

QUESTION $8$

There is no need to use tokenization when pretraining a BERT model.

True

False

QUESTION $9$

One of the techniques involved in BERT pretraining is Masked Language Modeling $($MLM$).$

True

False

QUESTION $10$

Transformer models should be compared using the same data set.

True

False