INDEX

Explanations

purpose-built anything

The neuron flags unusually long or uncommon tokens—often multi‐syllable, hyphenated, or domain‐specific words.

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

idjan

-1.61

大概是

-1.51

 says

-1.45

-1.44

%;

-1.43

 damos

-1.41

 AGAIN

-1.41

うち

-1.39

 sottile

-1.35

 makeover

-1.34

POSITIVE LOGITS

</strong>

1.61

 each

1.51

.”

1.48

);

1.48

:[{

1.34

();

1.33

 mármol

1.31

";

1.30

amping

1.30

Activations Density 0.037%