INDEX

Explanations

finishing;

The neuron fires on numeric tokens in reference citations—especially floating‐point numbers denoting journal volumes, issue pages, years, or similar bibliographic metadata.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 '');

0.89

 ',');

0.88

 "");

0.79

 "*");

0.72

})$;

0.70

이에요

0.68

()");

0.68

[]);

0.66

++);

0.66

 []);

0.63

POSITIVE LOGITS

2.58

2.30

;(

2.17

;-

2.09

;.

2.08

;')

2.06

;\

2.03

；

2.02

;,

1.98

;")

1.98

Activations Density 0.097%