INDEX

Explanations

of a, of the, of each

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

ä¸ĢäºĽ

-0.13

ä¸Ģäºº

-0.09

ares

-0.09

ä¸ĢçĤ¹

-0.09

ä¸ĢåĪĩ

-0.09

ä¸Ģå¹´

-0.09

 geography

-0.09

ä¸ĢåĢĭ

-0.09

ovah

-0.08

ä¸Ģæ¬¡

-0.08

POSITIVE LOGITS

 sorts

0.12

-buttons

0.11

/from

0.11

 surroundings

0.10

place

0.09

 riÃªng

0.09

ft

0.09

uez

0.09

 particular

0.08

 different

0.08

Activations Density 0.108%