INDEX

Explanations

choices and options

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

Specific

-0.11

 specific

-0.11

åħ·ä½ĵ

-0.11

specific

-0.10

 especÃŃf

-0.10

 specifically

-0.10

qid

-0.09

pecific

-0.09

 ÐºÐ¾Ð½ÐºÑĢÐµÑĤ

-0.09

 Ø®Ø§Øµ

-0.09

POSITIVE LOGITS

 popular

0.26

 Popular

0.23

popular

0.21

Popular

0.21

 choices

0.20

 options

0.20

choices

0.18

 Options

0.18

 possibilities

0.17

 Ð¿Ð¾Ð¿ÑĥÐ»ÑıÑĢ

0.17

Activations Density 0.081%