INDEX
Explanations
phrases indicating comparisons and contrasts
New Auto-Interp
Negative Logits
quite
-0.16
even
-0.16
entire
-0.15
both
-0.15
uela
-0.15
truly
-0.15
true
-0.15
directly
-0.15
iesel
-0.15
almost
-0.15
POSITIVE LOGITS
mere
0.30
mere
0.29
glor
0.26
bunch
0.22
merely
0.21
ãģŁãģł
0.20
COLLECTION
0.17
isol
0.17
inconvenience
0.17
Collection
0.17
Activations Density 0.294%