INDEX

Explanations

programming delimiters and keywords

The neuron strongly activates on the forward‐slash (“/”) token.

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

They

-1.88

</h4>

-1.80

There

-1.70

</h2>

-1.66

 Using

-1.64

Not

-1.54

has

-1.45

 Some

-1.44

</h5>

-1.42

just

-1.41

POSITIVE LOGITS

部分を

1.55

҃

1.48

\{

1.46

1.45

schnitt

1.41

⿷

1.41

熜

1.33

搧

1.33

玏

1.32

ίου

1.30

Activations Density 0.117%