INDEX

Explanations

can cause damage or failure

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

Glo

-0.10

Lak

-0.10

 leak

-0.10

 Glover

-0.09

 leaking

-0.09

 obstruct

-0.09

 patched

-0.09

 patch

-0.09

/weather

-0.09

forge

-0.09

POSITIVE LOGITS

 damage

0.18

 failure

0.16

 Damage

0.16

 safety

0.15

 heat

0.15

damage

0.14

Damage

0.14

 Failure

0.14

 overhe

0.13

 DAMAGE

0.13

Activations Density 0.066%