INDEX

Explanations

research topics and methods

The neuron detects mentions of research‐methodology terms (e.g. approaches, methods, investigations, questionnaires, monitoring routines).

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

itoba

-1.15

фов

-1.10

IONS

-1.10

 siè

-1.09

titch

-1.05

keted

-1.05

 ognuno

-1.03

ités

-1.03

amentals

-1.02

ità

-1.00

POSITIVE LOGITS

in

1.20

 such

1.17

ższych

1.11

 verändert

1.09

 frightening

1.08

or

1.05

 ці

1.05

’

1.02

}',

1.02

out

1.00

Activations Density 0.362%