INDEX

Explanations

The

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

 misunder

-0.24

ustomed

-0.23

 behalf

-0.23

å·²ä¹ħçļĦ

-0.23

 enthus

-0.23

atever

-0.22

ipop

-0.22

Sexy

-0.21

çļĦå¸¦é¢Ĩ

-0.21

echa

-0.21

POSITIVE LOGITS

oret

0.35

 reason

0.34

 goal

0.30

aim

0.29

çĽ®çļĦæĺ¯

0.29

 purpose

0.29

or

0.27

ories

0.26

 result

0.26

 only

0.25

Activations Density 1.842%