INDEX

Explanations

clearly or obviously followed by adjective/verb

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 arguably

-0.18

 surely

-0.17

 undoubtedly

-0.15

 actually

-0.15

 admittedly

-0.14

 supposedly

-0.13

 certainly

-0.13

 seemingly

-0.13

 invariably

-0.13

 reportedly

-0.13

POSITIVE LOGITS

SOM

0.10

 meant

0.10

 intended

0.09

 talented

0.09

 Haven

0.08

 pathname

0.08

 intent

0.08

ETS

0.08

 outlines

0.08

akk

0.08

Activations Density 0.068%