INDEX
Explanations
words related to suffixes indicating nationality or profession
New Auto-Interp
Negative Logits
ival
-0.17
ãĥ¥
-0.17
hybrid
-0.15
shoulder
-0.15
ivals
-0.15
DH
-0.14
Hybrid
-0.14
hybrids
-0.14
wo
-0.14
OTO
-0.14
POSITIVE LOGITS
éri
0.15
ssa
0.14
emoc
0.14
оÑĤов
0.14
itm
0.14
/schema
0.14
rij
0.14
raman
0.14
bak
0.14
_:
0.13
Activations Density 0.016%