INDEX

Explanations

using in code

The neuron selectively activates on numeric literal tokens (floating-point numbers) in code.

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

鹋

-1.74

our

-1.67

 geprü

-1.56

 drew

-1.55

zeppelin

-1.50

 bedeutend

-1.48

一旁的

-1.48

سيم

-1.48

 Schlafzimmer

-1.48

POSITIVE LOGITS

»;

1.72

with

1.67

↵↵

1.64

 certaine

1.38

ค์

1.36

inct

1.33

将她

1.33

енты

1.33

}}

1.32

ales

1.32

Activations Density 0.022%