INDEX
Explanations
specific references to historical documents or photographs, particularly from the American Civil War era
New Auto-Interp
Negative Logits
ubar
-0.17
basically
-0.17
pie
-0.16
orough
-0.15
ulpt
-0.15
abar
-0.15
-0.14
-0.14
subs
-0.14
acon
-0.14
POSITIVE LOGITS
taken
0.18
-caption
0.16
taken
0.16
Taken
0.16
frontal
0.16
Taken
0.16
negatives
0.15
caption
0.15
-Identifier
0.15
unidentified
0.15
Activations Density 0.067%