INDEX

Explanations

record safety incidents

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

lod

-0.09

Decision

-0.09

redd

-0.09

è°ĥæŁ¥

-0.08

åıĳèĤ²

-0.08

éĶº

-0.08

æĳĦåĥı

-0.08

ç¿¦

-0.08

è¤Ĭ

-0.08

_pas

-0.08

POSITIVE LOGITS

å®īåħ¨çĶŁäº§

0.23

å®īåħ¨ç®¡çĲĨ

0.22

 safety

0.18

å®īåħ¨

0.18

å®īåħ¨äºĭæķħ

0.17

çļĦå®īåħ¨

0.17

 Safety

0.17

Safety

0.17

äºĭæķħ

0.17

afety

0.15

Activations Density 0.096%