INDEX
Explanations
file paths
The neuron is triggered by tokens that look like file names or URLs—i.e. code or resource paths with familiar extensions (e.g. “.cc”, “.jar”, “.png”) or domain‐style strings.
New Auto-Interp
Negative Logits
比
-0.07
ведите
-0.07
finite
-0.06
ذكر
-0.06
rospy
-0.06
URRED
-0.06
thru
-0.06
RT
-0.06
FK
-0.06
_aliases
-0.06
POSITIVE LOGITS
وال
0.06
اورزی
0.06
')?>
0.06
_Code
0.06
мос
0.06
подраз
0.06
olmuş
0.06
đứ
0.06
BX
0.06
ковод
0.06
Activations Density 0.033%