INDEX
Explanations
significant structural and visual elements in descriptions, often focusing on buildings or landscapes
New Auto-Interp
Negative Logits
world
-0.17
erv
-0.15
ingers
-0.14
dealing
-0.14
ought
-0.14
ut
-0.13
cid
-0.13
raki
-0.13
prob
-0.13
Bast
-0.13
POSITIVE LOGITS
isch
0.15
ickers
0.15
ziel
0.14
oran
0.14
оÑĤо
0.14
ettel
0.14
į¼
0.14
SizePolicy
0.14
ains
0.14
eger
0.13
Activations Density 0.230%