INDEX

Explanations

references to torture and mistreatment in a prison context

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

 baiser

-0.08

chia

-0.08

etten

-0.07

icros

-0.07

opup

-0.07

assi

-0.07

pew

-0.07

etÃŃ

-0.07

ensi

-0.07

ultan

-0.07

POSITIVE LOGITS

 torture

0.08

 TORT

0.08

 Tort

0.07

 tort

0.07

 electric

0.07

ILD

0.06

 burns

0.06

 electrodes

0.06

priv

0.06

 interrog

0.06

Activations Density 0.030%