INDEX

Explanations

hypothetical scenarios

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

Hao

-0.10

rones

-0.09

 eventdata

-0.09

 astronom

-0.09

odyn

-0.09

Invoker

-0.09

incare

-0.08

-addons

-0.08

 injunction

-0.08

.wx

-0.08

POSITIVE LOGITS

AI

0.15

 Mars

0.13

AI

0.13

 world

0.12

 earth

0.12

 mars

0.12

 Earth

0.12

 consciousness

0.10

 person

0.10

æĻºèĥ½

0.10

Activations Density 0.350%