INDEX

Explanations

negative phrases indicating lack of authorization or need

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

Å¡ÃŃ

-0.08

dge

-0.07

wik

-0.07

 billig

-0.06

 odor

-0.06

/lab

-0.06

 paci

-0.06

Ã¨s

-0.06

iez

-0.06

krom

-0.06

POSITIVE LOGITS

 actually

0.12

 actual

0.10

actually

0.09

 ìĭ¤ìłľ

0.08

 Actual

0.08

 Actually

0.08

ibold

0.08

å®ŀéĻħ

0.08

Actually

0.07

çľŁæŃ£

0.07

Activations Density 0.024%