INDEX

Explanations

used to

The neuron fires on first-person, experience-reporting language (e.g. “I/me,” “worked for me,” “have used,” etc.), signaling personal anecdotes or testimonials.

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

 môžete

-1.50

to

-1.43

 opět

-1.39

 maximale

-1.32

 môžu

-1.27

 parfü

-1.27

presumably

-1.23

esterday

-1.22

 manuelle

-1.20

靭

-1.20

POSITIVE LOGITS

They

1.54

Very

1.45

 very

1.43

 each

1.39

That

1.37

 their

1.36

</h4>

1.35

so

1.32

Quite

1.31

Not

1.29

Activations Density 0.133%