INDEX

Explanations

Hend

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

Ø§Ø³Ø·Ø©

-0.12

 Ø§ÙĦØ±ÙħØ²ÙĬØ©

-0.11

'gc

-0.11

$MESS

-0.10

ruary

-0.10

 Abdullah

-0.10

âĦĸâĦĸ

-0.09

ò

-0.09

Ø±ÙĬÙĥÙĬØ©

-0.09

 Ø§ÙĦØ³Ø¹ÙĪØ¯ÙĬØ©

-0.09

POSITIVE LOGITS

al

0.12

ãĤĽ

0.12

Al

0.11

Ùį

0.09

 Greenwood

0.09

 vene

0.09

 merry

0.09

 Hend

0.09

Al

0.09

icher

0.09

Activations Density 0.083%