INDEX

Explanations

best possible

This neuron is primarily activated by content-bearing nouns and descriptive adjectives (e.g. “extreme,” “environments,” “technology,” “photographers,” “neighbors,” “people”)—in other words, key topic words.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 उपनिवेश

0.54

燾

0.53

 homologous

0.52

Gef

0.52

рока

0.52

婺

0.50

 Object

0.50

 dispensed

0.50

 যেটি

0.50

marshaller

0.50

POSITIVE LOGITS

 imaginable

1.76

 conceivable

1.31

 possible

1.28

possible

1.19

 Possible

1.16

 possíveis

1.13

Possible

1.09

 можли

1.09

 возмо

1.09

 ever

1.06

Activations Density 0.341%