INDEX

Explanations

effect or effects

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

ï½§

-0.11

 Falk

-0.09

Mor

-0.09

etat

-0.09

 Extras

-0.09

erca

-0.08

 Outlet

-0.08

atform

-0.08

 tang

-0.08

ippi

-0.08

POSITIVE LOGITS

 effect

0.59

 effects

0.54

 efect

0.47

 Effect

0.45

effect

0.43

 effet

0.43

 Effects

0.41

æķĪæŀľ

0.41

Effect

0.41

 ÑįÑĦ

0.39

Activations Density 0.162%