INDEX
Explanations
general snippets of information
The neuron responds to the opening words of major sentences or section headings—that is, it detects tokens at the very start of expository sentences.
New Auto-Interp
Negative Logits
lh
-0.07
ration
-0.07
_ascii
-0.07
افت
-0.06
orthodox
-0.06
backpack
-0.06
touching
-0.06
advising
-0.06
alex
-0.06
веч
-0.06
POSITIVE LOGITS
poměr
0.06
ου
0.06
IllegalArgumentException
0.06
:UIControl
0.06
.="<
0.06
_GT
0.06
("/:0.06
[token
0.06
.mo
0.06
碎
0.06
Activations Density 0.037%