INDEX
Explanations
mentions of specific dog breeds, especially “Retriever” and variations thereof
terms related to retro and nostalgic themes or references
New Auto-Interp
Negative Logits
inki
-0.81
ashtra
-0.73
Whitman
-0.66
jar
-0.65
veins
-0.59
cropped
-0.58
ullah
-0.58
mafia
-0.58
unrest
-0.57
instein
-0.57
POSITIVE LOGITS
dden
1.01
eval
0.85
Sax
0.82
©¶æ
0.76
heet
0.76
cycles
0.75
LECT
0.74
position
0.74
nels
0.72
POSE
0.71
Activations Density 0.063%