INDEX

Explanations

documentation formatting and markers

The neuron activates on numeric tokens (integers or decimals) in the text.

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

by

-1.79

It

-1.58

not

-1.51

as

-1.42

for

-1.32

 even

-1.32

 just

-1.27

 other

-1.27

so

-1.23

 first

-1.20

POSITIVE LOGITS

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.48

↵↵↵↵↵↵↵↵↵↵↵↵↵

1.42

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.39

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.39

↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.39

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.34

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.33

↵↵↵↵↵↵↵↵↵↵↵↵

1.32

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.27

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.27

Activations Density 0.005%