INDEX

Explanations

words starting with O and followed by specific suffixes

This neuron responds to structural formatting tokens—especially punctuation like colons, dashes, and numerals—rather than lexical content.

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

 anuncio

-0.76

懂

-0.73

 происходит

-0.70

extrémité

-0.69

冷凍

-0.68

孰

-0.68

 incluidas

-0.66

 движения

-0.66

 logro

-0.66

endroit

-0.65

POSITIVE LOGITS

1.68

 Olivia

1.44

 Odisha

1.40

 Oldham

1.39

 Oleg

1.38

オ

1.38

 Olivier

1.36

 octa

1.33

 Operand

1.33

 Owens

1.31

Activations Density 0.047%