INDEX

Explanations

unchanged or altered

This neuron is strongly activated by occurrences of the word “change” (including its variants like “Changed”).

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

InBackground

-1.27

 чем

-1.23

 nécessaire

-1.22

</sup>

-1.19

 partout

-1.16

tilor

-1.16

нении

-1.15

∬

-1.15

⅘

-1.13

</strong>

-1.13

POSITIVE LOGITS

 unchanged

1.53

 timmar

1.20

 vecka

1.17

 unaltered

1.17

 their

1.16

 hetzelfde

1.16

طان

1.15

 veckor

1.15

 except

1.15

 gånger

1.12

Activations Density 0.072%