INDEX

Explanations

(

The neuron fires on decimal numeric tokens (especially floating‐point numbers, often in parentheses).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

«,

0.74

(),

0.72

»,

0.71

»،

0.70

,»

0.70

】,

0.68

»,

0.68

rover

0.67

 —,

0.66

+}$,

0.65

POSITIVE LOGITS

Note

1.23

It

1.14

Although

1.14

Incidentally

1.13

There

1.08

This

1.05

Though

1.04

They

1.03

See

1.02

NOTE

1.02

Activations Density 0.068%