INDEX

Explanations

spatial coordinates

The neuron fires on words that mark locations or spatial relationships (e.g. prepositions like “in,” “inside,” “near,” and terms denoting places or adjacent areas).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 should

-1.55

 still

-1.38

 cannot

-1.27

 would

-1.19

になるので

-1.16

to

-1.13

 БО

-1.12

 that

-1.11

杳

-1.09

ꯢ

-1.09

POSITIVE LOGITS

it

1.46

ètre

1.24

 auspices

1.17

ड़

1.16

lysm

1.16

unno

1.13

ències

1.13

onation

1.11

the

1.09

 same

1.09

Activations Density 0.279%