INDEX

Explanations

to permit

The neuron consistently fires on occurrences of the word “to” immediately preceding a numeric value.

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

 této

-1.50

</h5>

-1.47

 trademark

-1.42

ﾒ

-1.40

螣

-1.37

 techniczne

-1.33

prüfung

-1.32

fee

-1.31

茕

-1.30

〗

-1.30

POSITIVE LOGITS

 these

2.64

 that

2.34

the

1.89

 those

1.84

 some

1.83

our

1.79

 this

1.79

 their

1.73

 with

1.63

 from

1.60

Activations Density 0.004%