INDEX

Explanations

focus on actions

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

 Opinion

-0.10

 Gotham

-0.09

ze

-0.09

 opinion

-0.09

ìĦĿ

-0.09

Ã¼t

-0.08

 opinions

-0.08

etto

-0.08

ç»Ĩ

-0.08

pip

-0.08

POSITIVE LOGITS

 focus

0.16

.focus

0.11

 focuses

0.11

focus

0.11

 Focus

0.11

_focus

0.10

 ÐºÐ¾Ð½ÑģÑĤÑĢÑĥÐº

0.10

 productive

0.09

oppins

0.09

neutral

0.09

Activations Density 0.065%