INDEX
Explanations
phrases and terms indicating completeness or fullness
New Auto-Interp
Negative Logits
ebi
-0.18
jte
-0.17
fullname
-0.17
jem
-0.17
ansen
-0.16
fullscreen
-0.16
FullScreen
-0.15
Fuller
-0.15
eful
-0.15
fullscreen
-0.15
POSITIVE LOGITS
blown
0.35
-scale
0.34
-bl
0.33
erton
0.33
fled
0.32
-length
0.32
scale
0.29
fil
0.29
eren
0.29
-on
0.27
Activations Density 0.071%