INDEX

Explanations

tax-deductible extent allowed

well-reasoned answer

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 trajets

0.51

 marqué

0.49

ajú

0.47

駭

0.47

 empfohlen

0.46

।-

0.46

 falso

0.46

 إلى

0.45

 incluem

0.45

 가정

0.45

POSITIVE LOGITS

 linen

0.46

nienie

0.44

 shell

0.43

これから

0.42

shell

0.41

 soon

0.41

 Paddy

0.41

modules

0.41

 ribbon

0.41

 Inspector

0.40

Activations Density 0.000%