INDEX

Explanations

a chiseled, curvy

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

 obese

-0.10

ubi

-0.10

 swath

-0.10

 Brilliant

-0.10

lid

-0.09

Fat

-0.09

 wizard

-0.09

stag

-0.09

 Prest

-0.09

 handsome

-0.09

POSITIVE LOGITS

fit

0.11

-fit

0.11

 beach

0.11

 iconic

0.11

Ã¶rper

0.10

 babel

0.10

InputDialog

0.10

arger

0.10

icon

0.09

abs

0.09

Activations Density 0.055%