INDEX

Explanations

do not have access

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

ãĢĢï½

-0.10

 Twist

-0.09

 Bannon

-0.09

à¥įà¤Ľ

-0.09

otu

-0.08

ogo

-0.08

bak

-0.08

INC

-0.08

abb

-0.08

ropa

-0.08

POSITIVE LOGITS

 access

0.28

 information

0.18

 enough

0.18

 knowledge

0.18

 acesso

0.17

access

0.15

 ability

0.14

 acceso

0.14

_access

0.13

knowledge

0.13

Activations Density 0.080%