INDEX
Explanations
descriptions of shapes and materials
New Auto-Interp
Negative Logits
anship
-0.98
arenthood
-0.95
utics
-0.91
osponsors
-0.87
illard
-0.84
Nation
-0.84
omics
-0.83
antics
-0.83
eligible
-0.83
eeks
-0.82
POSITIVE LOGITS
wooden
1.22
rectangular
1.15
staircase
1.15
cardboard
1.12
oval
1.12
rectangle
1.11
plastic
1.07
plaque
1.06
hole
1.06
ribbon
1.05
Activations Density 0.416%