INDEX

Explanations

references to risks and critical assessments in various contexts

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

 YardÄ±m

-0.09

 nackte

-0.09

ober

-0.09

edback

-0.08

eniz

-0.08

 seins

-0.08

kus

-0.07

adil

-0.07

TargetException

-0.07

Ø¨ÛĮÙĨÛĮ

-0.07

POSITIVE LOGITS

how

0.12

 whether

0.09

the

0.09

how

0.08

 ways

0.08

 cÃ³mo

0.07

 possible

0.07

 aspects

0.07

0.06

 aspect

0.06

Activations Density 0.027%