INDEX

Explanations

describing suffering or ailments

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

 sore

-0.10

 diarrhea

-0.10

 fever

-0.09

 cough

-0.09

tog

-0.09

 Hunger

-0.09

 cancers

-0.09

 emergencies

-0.09

 sickness

-0.09

assel

-0.08

POSITIVE LOGITS

 suffer

0.34

 suffering

0.30

 suffers

0.30

uffer

0.28

uffers

0.25

æĤ£

0.24

UFFER

0.22

 suffered

0.22

 Batt

0.16

 dealing

0.16

Activations Density 0.108%