INDEX
Explanations
news or video headlines with a sense of urgency or importance
the phrase "MUST WATCH" associated with video content
New Auto-Interp
Negative Logits
nowhere
-0.71
intent
-0.68
redevelopment
-0.63
lured
-0.62
scattering
-0.62
ura
-0.61
conver
-0.60
coales
-0.60
bowling
-0.60
recl
-0.60
POSITIVE LOGITS
VIDEOS
1.07
WATCH
1.00
Thumbnails
0.87
IMAGES
0.87
esome
0.83
WATCH
0.80
...]
0.77
gallery
0.73
WATCHED
0.72
!]
0.71
Activations Density 0.005%