INDEX

Explanations

questions and answers

The neuron selectively fires on tokens that signal questions or uncertainty—especially question marks and words like “but,” “because,” “of course,” and “sure” that introduce or emphasize doubt or queries.

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

лочка

-1.59

реи

-1.50

Oby

-1.47

mentes

-1.45

Kamera

-1.45

Struktur

-1.45

 maux

-1.41

 decidi

-1.41

 beij

-1.41

たちは

-1.40

POSITIVE LOGITS

and

3.00

 also

2.16

 with

1.80

in

1.78

on

1.77

1.67

 however

1.63

 since

1.62

 would

1.53

<blockquote>

1.43

Activations Density 0.081%