INDEX
Explanations
This neuron seems to be looking for a variety of words, but most strongly activates for the word "movies", so it is finding movie reviews
legal/technical documents
New Auto-Interp
Negative Logits
الحره
-0.57
celotti
-0.54
apunov
-0.52
>=",
-0.51
zano
-0.47
batim
-0.46
دریافتشده
-0.46
Awww
-0.46
masına
-0.45
<bos>
-0.44
POSITIVE LOGITS
ſeveral
0.70
Efq
0.68
ItemLayout
0.67
Reſ
0.65
Jefus
0.63
greateſt
0.62
CreateTagHelper
0.61
Majefty
0.60
Conſ
0.59
Monfieur
0.59
Activations Density 0.083%