INDEX
Explanations
disagreement and discussion
The neuron fires on tokens involved in first-person personal commentary or self-referential opinion (e.g. “I thought,” “I disagree,” “I reconsider”), i.e. authorial reflections.
New Auto-Interp
Negative Logits
(limit
-0.07
editorial
-0.07
goodies
-0.06
dữ
-0.06
Eaton
-0.06
(filter
-0.06
두
-0.06
↵ ↵
-0.06
лись
-0.06
result
-0.06
POSITIVE LOGITS
shocks
0.07
(boost
0.07
้ง
0.06
.Horizontal
0.06
наблю
0.06
ublice
0.06
.Multi
0.06
assignable
0.06
')?>
0.06
.empty
0.06
Activations Density 0.118%