INDEX
Explanations
conjunctions and transitional phrases indicating contrast or exception
New Auto-Interp
Negative Logits
isko
-0.19
itor
-0.17
Lad
-0.15
rika
-0.14
someone
-0.14
aghan
-0.14
odus
-0.14
ameron
-0.13
itia
-0.13
otle
-0.13
POSITIVE LOGITS
èĥĨ
0.17
elems
0.15
componentName
0.15
thanks
0.14
ymoon
0.14
ToOne
0.14
Ä
0.14
Hub
0.13
اÙĦÙħت
0.13
áž
0.13
Activations Density 0.105%