INDEX

Explanations

test and performance

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

åľ¨ä»Ĭå¹´

-0.27

åľ¨å¹¿å·ŀ

-0.24

 slipped

-0.24

æĪĳçĽ¸ä¿¡

-0.23

 TORT

-0.23

stre

-0.23

åľ¨å®¶éĩĮ

-0.23

 diplom

-0.23

_CAL

-0.23

elage

-0.22

POSITIVE LOGITS

è¿½æ±Ĥ

0.38

-driven

0.35

èĩ³ä¸Ĭ

0.33

ocracy

0.31

åĪ¶åº¦

0.31

 exig

0.29

å¯¼åĲĳ

0.28

 driven

0.28

 obsessed

0.27

æ³ķåĪĻ

0.27

Activations Density 0.053%