INDEX

Explanations

nÃ£o tem or having

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

shouldBe

-0.13

 shouldBe

-0.12

HasBeen

-0.11

.toBe

-0.10

CanBe

-0.10

 becomes

-0.09

 seront

-0.09

Been

-0.09

toBe

-0.09

 ÑģÑĤÐ°Ð½ÐµÑĤ

-0.09

POSITIVE LOGITS

 memiliki

0.40

æľīä¸Ģ

0.32

æľī

0.32

æĭ¥æľī

0.31

 having

0.31

 have

0.31

 Ð¸Ð¼ÐµÐµÑĤ

0.31

 Ð¸Ð¼ÐµÑİÑĤ

0.29

 avere

0.29

 Ð¸Ð¼ÐµÑĤÑĮ

0.28

Activations Density 0.399%