INDEX

Explanations

diploma work development planning

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

ÑĢÑı

-0.10

 hust

-0.09

 Couch

-0.09

~~

-0.09

gres

-0.09

flu

-0.09

 Hague

-0.09

alet

-0.08

Zub

-0.08

idel

-0.08

POSITIVE LOGITS

va

0.18

Ò

0.18

Ñŀ

0.16

Ò

0.16

0.15

 Ð¿ÑĢÐ¾ÑĦÐµÑģÑģÐ¸Ð¾Ð½Ð°Ð»ÑĮ

0.15

ning

0.13

deg

0.13

 assort

0.12

 ÑĦÐµÐ´ÐµÑĢÐ°Ð»ÑĮ

0.12

Activations Density 0.017%