INDEX
Explanations
questions and answers
This neuron detects occurrences of “data” (and related question words like need, accessible, collecting, testing) in the context of requirements‐style questions.
New Auto-Interp
Negative Logits
oker
-0.07
shortened
-0.07
Gate
-0.06
gy
-0.06
دارة
-0.06
sales
-0.06
också
-0.06
.neighbors
-0.06
받아
-0.06
weakened
-0.06
POSITIVE LOGITS
_background
0.07
SCI
0.07
background
0.06
concentr
0.06
(ball
0.06
vortex
0.06
�
0.06
convoy
0.06
undercut
0.06
диагности
0.06
Activations Density 0.075%