INDEX

Explanations

du experiencing illness

mentions of illness or physical symptoms (e.g., infections, swelling) in health-related contexts, including in non-English text.

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

Negative Logits

 bladder

-0.10

 Ð±Ð¾Ð»ÐµÐ·Ð½Ð¸

-0.10

çĻ

-0.09

 bowel

-0.09

 ÑĤÐ²Ð°

-0.09

 oscill

-0.09

Erd

-0.09

 Ð²Ð¿Ð¾Ð»

-0.09

 afflict

-0.09

 famine

-0.09

POSITIVE LOGITS

 feel

0.14

sym

0.14

 feels

0.13

 pain

0.13

çĹĽ

0.13

 felt

0.13

felt

0.12

 pains

0.12

 dolor

0.12

Ow

0.12

Activations Density 0.290%