INDEX

Explanations

level of abstract concepts

The neuron responds to words that denote quantitative measures or degrees—terms expressing levels, amounts, or magnitudes (e.g. “level,” “risk,” “ability,” “degree,” “progress,” “contamination”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 卧室

-1.60

 horloge

-1.47

 principalement

-1.46

 Municipalidad

-1.40

 randonnée

-1.40

 balade

-1.39

 strany

-1.38

 paille

-1.38

поді

-1.38

 moelle

-1.35

POSITIVE LOGITS

you

1.88

 that

1.51

1.50

1.45

1.43

 already

1.43

can

1.38

 will

1.34

 require

1.30

 использу

1.28

Activations Density 0.077%