INDEX

Explanations

Eureka! CNN Validity Tribal Warrior

This neuron activates on tokens inside programming code snippets—particularly the closing bracket of an array literal.

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

 perú

-1.27

もうすぐ

-1.23

 escritores

-1.18

 escote

-1.16

 incroy

-1.16

にほんブログ村

-1.16

を注文

-1.15

 achet

-1.14

 diminuer

-1.14

 bolsos

-1.13

POSITIVE LOGITS

1.51

to

1.50

1.48

on

1.28

1.27

1.22

ra

1.20

1.16

at

1.15

Activations Density 0.000%