INDEX

Explanations

display models like LCD, OLED

This neuron responds primarily to numeric tokens and digit‐containing identifiers (e.g. version numbers, model or part numbers, addresses).

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

-1.55

but

-1.51

so

-1.26

建設

-1.25

 expediente

-1.24

 progresses

-1.23

 помню

-1.22

此前

-1.21

 matrícula

-1.20

âce

-1.19

POSITIVE LOGITS

 otwor

1.39

 ladri

1.38

 zuży

1.34

تمبر

1.33

 réelle

1.30

{},

1.30

räumen

1.28

 odstran

1.27

Agora

1.25

틸

1.24

Activations Density 0.028%