INDEX

Explanations

This neuron fires on code tokens that mark data-structure field or property declarations—especially the field/property names and their type or attribute‐bracketing in struct/class definitions.

code structure

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 without

-1.27

 group

-1.15

 because

-1.11

 allow

-1.03

so

-1.03

 before

-1.02

 using

-1.02

 even

-1.01

 requires

-1.00

for

-0.97

POSITIVE LOGITS

Interes

1.28

또

1.11



1.06

 افتراض

1.05

 取り付け

1.04



1.04

 levier

1.03

Profes

1.03

 conster

1.02

甃

1.00

Activations Density 0.001%