INDEX

Explanations

exclusion from ban

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

<|begin_of_text|>

-0.14

******č\n

-0.12

¶Į

-0.12

įng

-0.12

ıa

-0.10

ÂĢÂĢ

-0.10

³Øª

-0.10

¨ë¶Ģ

-0.09

Ø§ÙĦØ¥ÙĨØ¬ÙĦÙĬØ²ÙĬØ©

-0.09

.Formatter

-0.09

POSITIVE LOGITS

 misunder

0.10

 proport

0.08

ss

0.07

ÐµÐºÐ¾

0.07

able

0.07

(using

0.07

 ë²łìĬ¤íĬ¸

0.07

 prive

0.06

 Freeman

0.06

Activations Density 0.721%