INDEX

Explanations

surface area and tension

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

ãģĬãĤĬ

-0.11

Hun

-0.10

 Falcon

-0.10

/tutorial

-0.09

icts

-0.09

venture

-0.09

ãģĦãģ¾ãģĻ

-0.09

zz

-0.09

ÄŁan

-0.09

der

-0.09

POSITIVE LOGITS

 tension

0.18

 area

0.13

-level

0.12

hog

0.12

.Surface

0.12

walker

0.11

-area

0.11

ICIAL

0.11

illance

0.11

area

0.11

Activations Density 0.019%