INDEX

Explanations

descriptive language

The neuron is detecting first‐person self‐references, i.e. occurrences of the pronoun “I.”

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

 skall

-1.26

的气息

-1.22

as

-1.19

Though

-1.18

Ambos

-1.15

Гц

-1.13

any

-1.10

kā

-1.09

 bereiken

-1.09

 incarnation

-1.08

POSITIVE LOGITS

 then

1.83

 разных

1.46

 чтобы

1.44

 kinda

1.41

 różne

1.41

so

1.37

 they

1.32

깐

1.30

 разные

1.28

 Obtén

1.23

Activations Density 0.084%