INDEX

Explanations

maze games and classic games

the neuron detects references to video games, especially iconic game titles and gameplay terms.

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

?”

-1.84

!”

-1.38

！”

-1.35

viens

-1.30

 eſſ

-1.27

thodox

-1.26

。”

-1.23

**/

-1.23

yards

-1.21

스트

-1.21

POSITIVE LOGITS

 związane

1.27

 szczeg

1.27

 other

1.25

 βρί

1.22

 apropiado

1.20

 reducido

1.14

摒

1.13

 select

1.12

 another

1.12

 there

1.10

Activations Density 0.014%