INDEX

Explanations

strategies/techniques

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

oa̍t

-0.86

 Jefus

-0.80

 houſe

-0.80

 becauſe

-0.78

 theſe

-0.77

Personensuche

-0.73

 ſeveral

-0.73

 itſelf

-0.73

 fubject

-0.71

 Monfieur

-0.71

POSITIVE LOGITS

WebVitals

0.61

of

0.49

 bicchiere

0.45

0.44

0.43

 вашего

0.42

 sommeil

0.42

ربعة

0.40

 cerimonia

0.40

Activations Density 0.134%