INDEX

Explanations

code syntax and structure

This neuron responds to standalone punctuation separator tokens—especially slashes ("/"), backslashes ("\"), and similar code‐ or markup‐style 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

こんな

-1.55

 всегда

-1.51

いつも

-1.38

というわけで

-1.34

Also

-1.34

For

-1.31

さてさて

-1.30

си

-1.27

 groote

-1.27

そんな

-1.27

POSITIVE LOGITS

animity

1.45

Poop

1.41

you

1.39

it

1.39

大満足

1.38

are

1.34

就去

1.33

 some

1.31

 they

1.30

 algunas

1.29

Activations Density 0.087%