Georgia tech is a big school of
Temperature

1

Embedding
Georgia
tech
is
a
big
school
of
Dropout
Residual
Layer Normalization
Multi-head Self Attention
Georgia
tech
is
a
big
school
of
Georgia
tech
is
a
big
school
of
Georgia
tech
is
a
big
school
of
QKT
Mask
Softmax
Attention
Head 1 of 12
MLP
Georgia
tech
is
a
big
school
of
Residual
11 more
Transformer
Blocks
go here.
Probabilities