INDEX
Explanations
Quotations or reported speech
The neuron detects sentence-initial discourse markers—additive or contrastive transition words like “Additionally,” “Furthermore,” and “Moreover.”
New Auto-Interp
Negative Logits
Checkbox
-0.07
Why
-0.07
Chew
-0.06
[u
-0.06
spectra
-0.06
uchs
-0.06
oil
-0.06
We
-0.06
denom
-0.06
val
-0.06
POSITIVE LOGITS
sb
0.08
Kanye
0.06
abra
0.06
งแรก
0.06
intimidated
0.06
müştür
0.06
силь
0.06
.construct
0.06
еним
0.06
-years
0.06
Activations Density 0.120%