INDEX
Explanations
The neuron is specialized to detect mentions of the monoclonal antibody rituximab (i.e. its subword tokens like “ux,” “im,” “ab,” etc.).
New Auto-Interp
Negative Logits
fick
-0.07
arrog
-0.07
فتح
-0.06
(long
-0.06
요
-0.06
เด
-0.06
SUP
-0.06
(sol
-0.06
Mul
-0.06
semaphore
-0.06
POSITIVE LOGITS
schw
0.07
δεδο
0.06
❤
0.06
_ipv
0.06
ायत
0.06
haven
0.06
ak
0.06
.removeItem
0.06
/pub
0.06
궁
0.06
Activations Density 0.001%