INDEX

Explanations

URLs and identifiers

The neuron fires strongly on short, punctuation‐delimited alphanumeric ID strings (e.g. random-looking codes or GUID fragments).

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

 while

-2.50

 without

-1.98

as

-1.94

 before

-1.90

 during

-1.90

 with

-1.88

at

-1.86

 would

-1.85

 only

-1.84

an

-1.83

POSITIVE LOGITS

킁

1.81



1.69

 チラ

1.55

鵙

1.55

 tivi

1.52

㈯

1.52

₶

1.51



1.51

 薬局

1.48

 kapture

1.48

Activations Density 0.042%