INDEX

Explanations

not have the ability to burn

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

Fir

-0.10

 Wagner

-0.10

 Amar

-0.09

 arsen

-0.09

Fee

-0.09

ca

-0.09

Fon

-0.09

ÑĥÐ»Ñİ

-0.09

 Extr

-0.09

ute

-0.08

POSITIVE LOGITS

 fuel

0.19

burn

0.18

 burn

0.18

çĩĥ

0.18

 burning

0.18

 comb

0.18

 chÃ¡y

0.17

sto

0.17

fuel

0.17

 Comb

0.16

Activations Density 0.039%