INDEX

Explanations

actions and inquiry

The neuron strongly activates on content words that denote actions—i.e. main lexical verbs (in both base and –ing forms) describing someone doing or starting something.

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

叽

-1.21

 cuándo

-1.16

 größer

-1.16

}$;

-1.15

 größeren

-1.12

):

-1.08

ेंगे

-1.07

);

-1.06

 Derfor

-1.05

という感じ

-1.05

POSITIVE LOGITS

myself

1.46

/>

1.30

 rouleau

1.21

 patrie

1.19

 myself

1.12

himself

1.10

1.07

 oeuvres

1.06

👀

1.05

hry

1.04

Activations Density 0.175%