INDEX

Explanations

ASCII, fetch, Font, default

The neuron strongly activates on isolated single‐character tokens (e.g. individual punctuation or symbol characters).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

of

-2.30

“

-1.70

 each

-1.66

 помню

-1.52

 belangrij

-1.49

at

-1.48

to

-1.45

on

-1.45

 chère

-1.43

 \%}$

-1.41

POSITIVE LOGITS

ICOLON

1.60

狁

1.59

."),

1.53

了

1.51

sandalia

1.49

 líqu

1.48

還有

1.44

 новые

1.43

 halben

1.42

短暂

1.42

Activations Density 0.072%