INDEX

Explanations

development, health, and dreams

The neuron is triggered by long, multisyllabic content words (e.g. “historic,” “character,” “arrhythmias,” “formulation,” “molding”).

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

for

-1.66

in

-1.59

魮

-1.35

᎑

-1.30

“...

-1.28

:['

-1.20

tems

-1.20

ஂ

-1.15

tidak

-1.14

 habilidades

-1.14

POSITIVE LOGITS

</strong>

1.60

our

1.30

BDL

1.30

нович

1.23

ciaż

1.22

},

1.22

its

1.21

ropical

1.20

can

1.19

 Utiliz

1.19

Activations Density 0.100%