INDEX

Explanations

making statements or comments

The neuron activates on reporting and attribution language—words like “said,” “claimed,” “comments,” “remarks,” “interview,” or “made” that introduce or describe quoted or reported speech.

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

 from

-1.72

 Already

-1.52

Jäh

-1.40

 nuestra

-1.38

襖

-1.30

 виправи

-1.29

 unsere

-1.29

of

-1.27

捺

-1.26

aktery

-1.25

POSITIVE LOGITS

 about

1.38

饜

1.28

が悪い

1.27

obf

1.23

 компьютера

1.19

 kwar

1.16

憮

1.16

 afric

1.16

 programu

1.15

 ernst

1.14

Activations Density 0.130%