INDEX

Explanations

first

The neuron selectively fires on words naming website‐ or software‐interface components and design/layout terms (e.g. “layout,” “header,” “desktop,” “version,” “search boxes,” “site”).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

ण्याच्या

0.52

ተር

0.50

czna

0.48

attoo

0.48

ahirkan

0.47

 mọi

0.46

 করিলে

0.46

រក

0.46

 erupt

0.45

<unused20>

0.45

POSITIVE LOGITS

 first

4.78

first

4.50

 firstly

4.34

首先

4.22

 primero

4.13

 først

4.13

 zuerst

4.03

 먼저

3.97

Firstly

3.95

 primeiro

3.89

Activations Density 1.030%