INDEX

Explanations

telling stories or events

This neuron activates on structural or metadata markers in a document—chiefly section headings, scene labels, and other numbered/enumerated formatting (e.g. “Act 4,” “Scene 1,” numbered items, and related punctuation).

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

 ſtre

-2.08

ֶּ

-1.91

ּוֹ

-1.88

 😁

-1.88

 kilometer

-1.88

菒

-1.85

 effectuer

-1.83

ִּ

-1.82

 Dinas

-1.82

 unbelievably

-1.80

POSITIVE LOGITS

 after

2.69

 with

2.23

one

2.19

 during

2.17

on

2.13

at

2.05

all

1.98

 emphasizing

1.95

“

1.94

 without

1.94

Activations Density 0.034%