INDEX

Explanations

body and its associated content

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

ANGO

-0.11

ally

-0.10

(s

-0.10

mast

-0.10

ery

-0.10

lz

-0.10

sWith

-0.09

sto

-0.09

 personalities

-0.09

(strtolower

-0.09

POSITIVE LOGITS

guards

0.22

guard

0.21

 politic

0.21

weight

0.17

ÑħÑĢÐ°Ð½

0.16

é¨ĵ

0.16

builder

0.14

work

0.14

éªĮ

0.14

ied

0.13

Activations Density 0.039%