INDEX

Explanations

go back

The neuron strongly activates on very high‐frequency common tokens—especially function words like pronouns, auxiliary verbs, prepositions, and conjunctions.

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

</h3>

-1.30

 durer

-1.19

The

-1.19

 colheres

-1.16

ーン

-1.16

 Основ

-1.13

ДУ

-1.13

 étro

-1.13

yasi

-1.12

сп

-1.11

POSITIVE LOGITS

 those

1.34

 peligroso

1.30

 even

1.27

그러나

1.26

 extraordinario

1.17

 چربی

1.16

Ой

1.13

🐹

1.11

 confiable

1.09

now

1.08

Activations Density 0.044%