INDEX
Explanations
proper nouns, particularly place names and street addresses
New Auto-Interp
Negative Logits
atron
-0.17
BATCH
-0.17
اÙĨÙĬØ©
-0.15
enson
-0.15
consect
-0.15
éĦī
-0.14
367
-0.14
penetration
-0.14
agna
-0.14
.si
-0.13
POSITIVE LOGITS
Invent
0.15
allax
0.15
arton
0.14
/Dk
0.14
ittel
0.14
/left
0.14
Koh
0.13
prefer
0.13
Og
0.13
/right
0.13
Activations Density 0.142%