INDEX
Explanations
weekly/monthly reports
The neuron fires on those framing/lede sentences that give a periodic summary or update—especially ones using temporal cues like “this week,” “finally,” and similar overview-style phrasing.
New Auto-Interp
Negative Logits
variant
-0.07
ст
-0.07
Jak
-0.06
and
-0.06
IQ
-0.06
933
-0.06
recognize
-0.06
ophile
-0.06
680
-0.06
Name
-0.06
POSITIVE LOGITS
frenzy
0.07
\param
0.07
)]↵↵
0.06
лись
0.06
QPointF
0.06
omidou
0.06
het
0.06
знач
0.06
Began
0.06
.gallery
0.06
Activations Density 0.064%