INDEX

Explanations

auxiliary verbs and modals

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

******č\n

-0.12

¶Į

-0.12

ÂĢÂĢ

-0.11

¦æĥħ

-0.11

įng

-0.11

EMPLARY

-0.10

/null

-0.10

oleÄį

-0.09

 -*-č\n

-0.09

 yoksa

-0.09

POSITIVE LOGITS

 itself

0.21

/is

0.15

its

0.14

Its

0.11

its

0.10

 Walters

0.10

Its

0.10

or

0.09

 Wend

0.09

Activations Density 0.511%