INDEX

Explanations

very or a lot followed by word

The neuron detects degree-intensifying words (e.g. “very,” “long,” “great”) that amplify descriptions.

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

-1.48

 such

-1.45

-1.34

mani

-1.30

Instead

-1.29

 plötzlich

-1.28

^=

-1.27

 that

-1.27

-1.26

POSITIVE LOGITS

 also

1.59

 arī

1.58

}$,

1.52

 pokud

1.45

 pelindung

1.39

 myös

1.38

 cepillo

1.36

ových

1.34

šanas

1.34

 ugyan

1.32

Activations Density 0.084%