INDEX

Explanations

nodded or told

The neuron is keyed to the opening quotation mark that signals the start of a spoken‐dialogue segment.

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

who

-1.34

</h4>

-1.34

-1.30

Before

-1.24

くんの

-1.21

Какой

-1.20

How

-1.15

he

-1.15

how

-1.13

重点

-1.12

POSITIVE LOGITS

to

1.40

Depends

1.32

OOOO

1.30

 cend

1.18

 Fichier

1.17



1.16

veloped

1.15

ZZZ

1.15

 Seeder

1.14

OOOOO

1.13

Activations Density 0.019%