INDEX

Explanations

supportive roles and services

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

gow

-0.10

elop

-0.10

gib

-0.10

pite

-0.09

lendi

-0.09

stin

-0.09

ibur

-0.09

esser

-0.09

vara

-0.09

ouver

-0.09

POSITIVE LOGITS

ive

0.28

ively

0.25

iveness

0.23

IVE

0.18

ingly

0.16

ability

0.15

ByVersion

0.14

able

0.13

ably

0.13

\Facades

0.12

Activations Density 0.028%