INDEX

Explanations

division, ND, BL, TI, violation

The neuron is strongly activated by isolated ampersand (“&”) tokens (and the adjacent single-letter “S” tokens) used in the slash‐separated address fields of email headers.

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

-1.88

 gets

-1.62

has

-1.62

on

-1.61

my

-1.59

 either

-1.56

so

-1.53

 enhances

-1.52

or

-1.50

 with

-1.49

POSITIVE LOGITS

搋

1.99

(“

1.99

 pleaf

1.98

 caufe

1.96

 occaf

1.95

艄

1.94

 moft

1.92

歺

1.92

>");

1.91

⚌

1.91

Activations Density 0.052%