INDEX

Explanations

added unique character

The neuron activates on marketing-style nouns that signal a unique or distinctive feature—words like “twist,” “edge,” “touch,” and “character.”

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

-1.74

like

-1.69

 when

-1.55

 there

-1.54

le

-1.38

-1.37

 from

-1.34

 only

-1.34

nd

-1.29

-1.28

POSITIVE LOGITS

LEncoder

1.57

xase

1.51

âmara

1.49

нева

1.48

𓄹

1.47

Quién

1.42

πω

1.41

$,

1.41

kać

1.39

providedIn

1.38

Activations Density 0.035%