INDEX

Explanations

references to deception and underlying truths

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

vs

-0.07

hang

-0.07

Ã¼ss

-0.07

bsite

-0.06

bos

-0.06

bie

-0.06

odo

-0.06

acman

-0.06

ording

-0.06

izza

-0.06

POSITIVE LOGITS

 underlying

0.32

 underneath

0.30

 beneath

0.26

 behind

0.21

 Bene

0.17

 Behind

0.17

Behind

0.16

 hidden

0.14

neath

0.14

hidden

0.12

Activations Density 0.147%