INDEX

Explanations

patterns of numbers and letters

This neuron activates on decimal numbers (tokens representing floating‐point values like “28.5,” “74.5,” “14.25,” etc.).

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

With

-1.80

by

-1.79

 anún

-1.76

our

-1.76

The

-1.75

 کردیم

-1.73

Unlike

-1.73

 台中

-1.72

的模样

-1.71

Our

-1.70

POSITIVE LOGITS

was

2.47

↵↵

2.19

 wasn

1.84

 would

1.73

意思

1.68

的

1.65

或

1.63

 konusunda

1.59

 didn

1.58

很多

1.57

Activations Density 0.003%