INDEX

Explanations

Bit compounds and URLs

This neuron specializes in detecting the token “bit” (in any casing) wherever it appears.

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

灕

-2.88

 土産

-2.55

愎

-2.50

 himſelf

-2.47

鶿

-2.45

vantaged

-2.38

 siano

-2.38

蜑

-2.34

𐐀

-2.33

 одном

-2.31

POSITIVE LOGITS

 необходимо

3.30

</i>

3.30

</h1>

3.02

2.78

饪

2.75

 Bereichen

2.69

2.63

At

2.52

entikan

2.45

And

2.44

Activations Density 0.010%