INDEX

Explanations

looking around or at

The neuron activates on words denoting visual attention (e.g. look, looked, looking, glance).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

we

-1.30

はこの

-1.24

astricht

-1.18

gdy

-1.16

 when

-1.15

 Untold

-1.14

 Präsidenten

-1.13

nología

-1.13

))

-1.12

นี้

-1.11

POSITIVE LOGITS

 away

1.19

Finally

1.13

 mahu

1.12

1.11

ongono

1.09

 природа

1.08

 vernac

1.06

 kili

1.05

 briefly

1.05

)$;

1.05

Activations Density 0.005%