INDEX
Explanations
Locations
This neuron fires on numeric measurement tokens (including decimal numbers) in technical descriptions.
New Auto-Interp
Negative Logits
unta
-0.07
Development
-0.07
justice
-0.07
fees
-0.06
chers
-0.06
ман
-0.06
MouseListener
-0.06
USD
-0.06
stery
-0.06
constant
-0.06
POSITIVE LOGITS
charm
0.07
thì
0.07
expand
0.07
|-
0.07
ном
0.07
_live
0.07
,存于
0.07
.comments
0.06
silently
0.06
.getRight
0.06
Activations Density 0.030%