INDEX
Explanations
words related to wind, such as "windfall" or "windscreen"
New Auto-Interp
Negative Logits
UNCH
-0.66
onse
-0.66
acho
-0.64
uder
-0.64
Lonely
-0.63
xus
-0.63
henko
-0.62
usalem
-0.62
earch
-0.61
endants
-0.61
POSITIVE LOGITS
shield
1.25
mill
1.14
turbines
1.06
turbine
1.01
lass
1.00
blown
0.97
lasses
0.94
bow
0.93
fall
0.89
walker
0.87
Activations Density 0.020%