INDEX

Explanations

numbers and code fragments

The neuron strongly activates on individual numeric characters (digits) 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

or

-1.65

but

-1.55

 riuscito

-1.51

?";

-1.43

 новом

-1.39

and

-1.39

🩷

-1.37

thc

-1.32

 یا

-1.31

=="

-1.28

POSITIVE LOGITS

 There

1.70

And

1.66

 Therefore

1.57

It

1.55

 Their

1.51

 Umfang

1.48

 Urteil

1.44

さんで

1.42

 That

1.42

How

1.42

Activations Density 0.026%