INDEX

Explanations

doing or performing well/badly

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

illac

-0.09

 Mild

-0.09

 Dome

-0.09

igest

-0.08

 Partial

-0.08

IFS

-0.08

estroy

-0.08

 partial

-0.08

ful

-0.08

upon

-0.08

POSITIVE LOGITS

 poorly

0.28

 well

0.24

 better

0.23

better

0.19

well

0.18

 WELL

0.18

 badly

0.17

 Better

0.17

 mieux

0.17

 excell

0.16

Activations Density 0.151%