INDEX

Explanations

starting word or punctuation

This neuron activates strongly on content-carrying tokens (nouns, pronouns, and key symbols) rather than common function words.

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

ensed

-1.50

*$

-1.43

³)

-1.34

$,

-1.33

(!__

-1.30

obé

-1.29

🕗

-1.28

</h1>

-1.27

}:=

-1.27

CORBA

-1.26

POSITIVE LOGITS

 assuring

1.29

去年

1.23

 cautiously

1.22

 osserva

1.22

 noastre

1.21

 aggiunge

1.20

ных

1.19

AMENTO

1.18

res

1.16

 respectivas

1.16

Activations Density 0.000%