INDEX
Explanations
references to various types of soda and beverages
New Auto-Interp
Negative Logits
heed
-0.90
ioned
-0.73
otos
-0.70
ammad
-0.69
OST
-0.69
isations
-0.69
aldo
-0.68
ebus
-0.68
akens
-0.68
inel
-0.67
POSITIVE LOGITS
bottles
0.88
cans
0.85
cups
0.84
bottle
0.82
drinkers
0.82
soda
0.81
dispens
0.79
refill
0.78
Bottle
0.77
Stream
0.76
Activations Density 0.007%