INDEX

Explanations

but followed by failure

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 also

-1.05

 further

-0.96

 another

-0.91

MENAFN

-0.91

چنین

-0.91

参戦

-0.89

 more

-0.89

 Strategy

-0.88

ジング

-0.88

izowane

-0.88

POSITIVE LOGITS

 siempre

1.45

 never

1.39

 always

1.39

Ultimately

1.32

Finally

1.30

 nothing

1.30

 sempre

1.28

 תמיד

1.27

NONE

1.25

 each

1.22

Activations Density 0.128%