INDEX

Explanations

RSA private key

The neuron strongly activates on short sequences of alphabetic characters of two or three letters (e.g. “AC,” “AA,” “AK,” “AG,” “ATA”) embedded in code‐ or data‐like contexts.

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

and

-1.93

-1.82

or

-1.70

</h1>

-1.45

It

-1.43

-1.39

-1.38

 Then

-1.35

My

-1.34

 toont

-1.34

POSITIVE LOGITS

ління

1.68

 HAZARD

1.67

utiliser

1.63

的支持

1.59

 ApiException

1.59

옻

1.58

esus

1.58

 setPassword

1.55

 useAppContext

1.55

 BaseController

1.52

Activations Density 0.001%