INDEX

Explanations

code and file paths

This neuron activates on individual hexadecimal characters (0–9 and a–f) that appear as parts of URL fragment identifiers or similar anchor‐ID strings.

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

 его

-1.38

,"

-1.37

婶

-1.30

好不好

-1.30

),"

-1.30

着她

-1.28

吀

-1.28

日まで

-1.25

тье

-1.24

bois

-1.24

POSITIVE LOGITS

1.39

↵

1.18

？』

1.17

1.15

 contactez

1.14

౭

1.14

SetColor

1.13

utilité

1.13

 Compromise

1.11

',

1.09

Activations Density 0.003%