INDEX

Explanations

This neuron specializes in spotting structural or markup‐style tokens—like code identifiers, JSON keys/values, table headings, scenario or test labels, and other technical‐document markers.

citations like [@CIT0001]

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

 enää

-1.07

 elétrica

-1.03

盍

-0.98

ண்ண

-0.97

 bricola

-0.96

 hyö

-0.95

 eivät

-0.91

⧠

-0.91

 unicórnio

-0.91

 löytyy

-0.90

POSITIVE LOGITS

janje

1.06

 trong

0.81

𝙽

0.79

全

0.79

是怎么回事

0.77

nasio

0.77

av

0.77

фф

0.73

 вжи

0.73

 intra

0.71

Activations Density 0.012%