INDEX

Explanations

indexed sequences starting with 1

The neuron activates on the digit “1,” especially when it appears in mathematical expressions or as a subscript.

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

Despite

-2.69

 Before

-2.66

They

-2.52

Although

-2.52

Surprisingly

-2.48

This

-2.44

During

-2.38

Though

-2.33

 Following

-2.30

Before

-2.27

POSITIVE LOGITS

箧

2.70

 travailler

2.64

⚣

2.53

蠖

2.39

鋱

2.33

2.31



2.27

 propriétés

2.25

鳎

2.25

 décider

2.23

Activations Density 0.008%