INDEX

Explanations

urls and domains

The neuron detects pieces of network addresses—e.g. the “@” in email addresses, domain-name tokens like “.com,” and URL separators.

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

✨:

-1.70

*</

-1.66

ternut

-1.65

hever

-1.64

lllll

-1.63

<bos>

-1.63

:{},

-1.60

 guantes

-1.59

 három

-1.57

 transportasi

-1.54

POSITIVE LOGITS

to

2.22

or

1.61

 being

1.52

 already

1.50

in

1.48

 there

1.45

 again

1.45

 else

1.42

1.41

so

1.35

Activations Density 0.003%