INDEX

Explanations

woes, troubles, and problems

The neuron flags nouns that denote troubles or difficulties (e.g., “problems,” “woes,” “arrest”).

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

旆

-1.66

 such

-1.59

the

-1.45

tween

-1.41

 Rektor

-1.37

#%%

-1.30

 {}));

-1.28

‗

-1.28

 wann

-1.27

áctica

-1.27

POSITIVE LOGITS

 tienen

1.49

1.47

 сча

1.45

とりあえず

1.41

相変わらず

1.32

 Mientras

1.31

 woes

1.30

Según

1.30

Mientras

1.28

1.24

Activations Density 0.049%