INDEX
Explanations
The neuron detects the standard “historic … located at …” geographic‐location phrasing (e.g. “is a historic … building located at …”) in building descriptions.
New Auto-Interp
Negative Logits
_sleep
-0.07
賢
-0.06
criticize
-0.06
izer
-0.06
defin
-0.06
「そう
-0.06
щи
-0.06
먹
-0.06
PLAIN
-0.06
/******************************************************************************/↵
-0.06
POSITIVE LOGITS
Penis
0.07
embodies
0.07
Produto
0.07
Ав
0.07
embodied
0.07
horn
0.06
атор
0.06
(Long
0.06
stands
0.06
ACE
0.06
Activations Density 0.005%