INDEX

Explanations

Ukrainian defense and authorities

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

 Rebellion

-0.11

 retries

-0.10

âī¦

-0.10

 compass

-0.09

 Compass

-0.09

 whitespace

-0.09

vero

-0.09

 silenced

-0.09

omentum

-0.08

 uncont

-0.08

POSITIVE LOGITS

 defense

0.37

 defence

0.33

 defend

0.32

 defending

0.30

 defenders

0.27

 Ø¯ÙģØ§Ø¹

0.27

defense

0.27

 Defense

0.25

-defense

0.25

 defended

0.25

Activations Density 0.092%