INDEX

Explanations

A

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Ressource

-0.66

الحياه

-0.66

 Ressources

-0.65

 الاطلاع

-0.65

__':

-0.63

 propOrder

-0.62

numerusform

-0.60

ARG

-0.60

كويكب

-0.59

awtextra

-0.59

POSITIVE LOGITS

featureID

0.49

0.48

 &___

0.40

 surla

0.40

stage

0.39

 mounted

0.38

retmen

0.38

mators

0.37

zdjęcie

0.36

 utafitiHapana

0.36

Activations Density 0.021%