INDEX

Explanations

hyperlinks and tabs

This neuron fires on words referring to navigational or link‐related UI elements (e.g. “blue,” “links,” “tabs,” “linked”).

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

 these

-1.70

and

-1.70

 ersten

-1.64

⻬

-1.59

猢

-1.38

 abarca

-1.37

 ordenado

-1.36

🤛

-1.34

 primeiros

-1.33

戗

-1.33

POSITIVE LOGITS

2.00

</strong>

1.65

romolecules

1.58

ஂ

1.55

tives

1.55

ي

1.55

 ")";

1.51

not

1.50

using

1.50

られる

1.48

Activations Density 0.001%