INDEX
Explanations
Android code and API responses
The neuron strongly activates on programming‐style tokens—e.g. variable names, assignment operators, and string literals in code snippets.
New Auto-Interp
Negative Logits
_memory
-0.07
культур
-0.06
-*-↵↵
-0.06
ries
-0.06
izons
-0.06
_Node
-0.06
wool
-0.06
velop
-0.06
.PUT
-0.06
勝
-0.06
POSITIVE LOGITS
$sub
0.07
<?
0.07
황
0.07
lis
0.07
TRANS
0.06
fishing
0.06
прий
0.06
SQ
0.06
ibili
0.06
.Serial
0.06
Activations Density 0.023%