INDEX
Explanations
phrases indicated by a special character sequence "***"
repeated special characters or symbols
New Auto-Interp
Negative Logits
subcommittee
-0.70
vation
-0.69
utive
-0.68
scattering
-0.68
curv
-0.67
etheless
-0.66
exting
-0.65
gaze
-0.64
scope
-0.64
foc
-0.64
POSITIVE LOGITS
NEW
0.86
Edited
0.83
!/
0.82
edited
0.81
WARNING
0.80
***
0.79
EDIT
0.79
***
0.79
TOP
0.77
THIS
0.75
Activations Density 0.019%