INDEX

Explanations

printf

The neuron fires on tokens that are numeric literals (i.e. any token containing digits).

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

</h2>

-1.77

its

-1.75

</h4>

-1.64

how

-1.59

or

-1.55

-1.51

It

-1.51

and

-1.50

 develop

-1.49

it

-1.47

POSITIVE LOGITS

雳

2.13

慥

1.89

łgorzata

1.88

 seks

1.79

 houſe

1.78

 ſtand

1.78

 ſou

1.77

 viſ

1.75

 kedi

1.73

 ſtate

1.73

Activations Density 0.003%