INDEX
Explanations
.org websites
The neuron activates on pieces of URLs (e.g. “http://”, domain names, “.org”, “/” path separators), so it’s looking for URL or web‐address fragments.
New Auto-Interp
Negative Logits
鑑
-0.07
starter
-0.07
_hid
-0.06
plastics
-0.06
trainers
-0.06
-d
-0.06
Aboriginal
-0.06
training
-0.06
animate
-0.06
({-0.06
POSITIVE LOGITS
дней
0.06
backs
0.06
اق
0.06
çık
0.06
düşür
0.06
ление
0.06
cooperate
0.06
başka
0.06
kino
0.06
find
0.06
Activations Density 0.013%