INDEX

Explanations

all data and values

The neuron flags technical nouns referring to information artifacts—things like data, files, patterns, updates, values, and results.

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

in

-2.11

by

-2.08

on

-1.77

 there

-1.49

any

-1.48

as

-1.43

 through

-1.41

 even

-1.35

at

-1.27

 making

-1.19

POSITIVE LOGITS

Either

1.49

扨

1.38

 女の子

1.34

how

1.32

 insegu

1.30

 Either

1.27

щению

1.27

فقط

1.24

Уважа

1.23

 الذي

1.22

Activations Density 0.217%