INDEX

Explanations

self-awareness, situational awareness, awareness of

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

Not in Any Lists

No Comments

Negative Logits

ServletResponse

-0.10

 Gors

-0.10

çĥĪ

-0.10

ews

-0.09

ç±į

-0.09

burgh

-0.09

icter

-0.09

ellers

-0.09

entai

-0.09

ä»Ķ

-0.09

POSITIVE LOGITS

es

0.21

ness

0.20

0.19

-ra

0.15

nes

0.14

aware

0.12

-aware

0.12

 aware

0.12

NESS

0.12

 awareness

0.12

Activations Density 0.017%