INDEX

Explanations

references to personal values and moral judgments

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

icket

-0.07

ayo

-0.07

ox

-0.07

lea

-0.06

 Fare

-0.06

oon

-0.06

anko

-0.06

noch

-0.06

ooke

-0.06

indy

-0.06

POSITIVE LOGITS

 ought

0.13

 shouldn

0.12

 should

0.12

should

0.11

 deserves

0.10

 surely

0.09

 must

0.09

 deserve

0.09

Should

0.09

 sollten

0.09

Activations Density 0.045%