INDEX

Explanations

operating, conducting, performing actions

New Auto-Interp

Top Features by Cosine Similarity

Configuration

Prompts (Dashboard)

10,000 prompts, 128 tokens each

Dataset (Dashboard)

lmsys/lmsys-chat-1m

Embeds

IFrame

Link

Negative Logits

-0.10

 Willis

-0.09

 Äįlov

-0.09

aggi

-0.09

ungal

-0.08

untime

-0.08

_POLL

-0.08

 Lazar

-0.08

ldre

-0.08

ongo

-0.08

POSITIVE LOGITS

 operating

0.24

 operate

0.20

 Operating

0.19

 operates

0.19

Operating

0.17

 operated

0.16

 conducting

0.16

operate

0.16

æĺ¯åľ¨

0.16

 Conduct

0.15

Activations Density 0.231%

No Known Activations