INDEX

Explanations

The incident

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

âĹİ

-0.09

 AraÅŁtÄ±r

-0.08

 regular

-0.08

forge

-0.08

 spare

-0.08

abase

-0.08

 Sunder

-0.08

 Target

-0.08

 Ventures

-0.08

CCI

-0.08

POSITIVE LOGITS

 incident

0.24

incident

0.18

 exact

0.17

ories

0.15

 incidents

0.14

 identities

0.14

 episode

0.13

odor

0.13

 event

0.13

 matter

0.12

Activations Density 0.060%