INDEX

Explanations

highly

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

afficheront

-0.74

StructEnd

-0.73

featureID

-0.71

évaluateur

-0.71

 disambiguazione

-0.71

protoimpl

-0.68

 esternos

-0.68

ंदीखरीदारी

-0.66

 chi̍t

-0.66

Jeografia

-0.66

POSITIVE LOGITS

 regarded

0.61

 scores

0.52

 raids

0.51

 respected

0.50

 ratings

0.48

 anticipated

0.47

 scoring

0.44

 praised

0.44

 commended

0.43

 accolades

0.43

Activations Density 0.016%