INDEX

Explanations

professional and trustworthy

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

èľľ

-0.10

Absent

-0.09

oppel

-0.09

 reconnaissance

-0.09

 Fork

-0.09

illon

-0.08

 Imports

-0.08

 cant

-0.08

bower

-0.08

ucch

-0.08

POSITIVE LOGITS

 trust

0.14

 serious

0.14

trust

0.14

 Trust

0.12

 Serious

0.11

 professional

0.11

 sophistic

0.11

 Authority

0.11

 authority

0.11

 modern

0.10

Activations Density 0.073%