INDEX

Explanations

know how to

This neuron detects instances of the verb “know” when the text asserts knowledge (e.g. “I know…,” “know how to…,” “know why…”).

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

-1.31

-1.23

 itself

-1.21

ついで

-1.20

 воспользоваться

-1.17

 lagen

-1.14

 собрать

-1.14

 eram

-1.14

“

-1.13

 hasn

-1.11

POSITIVE LOGITS

but

1.73

وقال

1.38

ův

1.34

 följande

1.30

 प्रत्येक

1.26

 varandra

1.23

 Mitteln

1.22

 pewno

1.21

$;

1.20

sə

1.20

Activations Density 0.053%